A common view is that each individual human may be tempted to cheat on a social dilemma, even if compliance by all would be a net benefit to all. What if they were all to punish a cheater? The same dilemma is replicated: each is tempted to refrain from costly punishment, even though punishment by all would be a net benefit to all. And so on, in infinite regress. Enforcement of norms of several kinds, not just social dilemmas, is a problem for egoistic rational-choice theory. One of the economists’ favorite maneuvers is to suggest that the people involved will interact repeatedly into the indefinite future, and thus that each, in order to pocket gains from those future interactions, will be egoistically and instrumentally motivated to maintain a reputation. Instrumental egoism is a far too narrow and dogmatic account of human motivation, however.
Christine Horne identifies three families of theories about norms. The consequentialist holds that the greater the benefits of norm compliance, the more likely is effective norm enforcement. The shared-meanings theorist attributes norm compliance to agreement on appropriate action. The typicality theorist attributes norm compliance to individual motivation to imitate population-typical behaviors.
Horne adds a fourth family, dubbed the relational theory, that would provide an explanation for some instances of enforcement of norms (she defines norms as rules, agreed on by many, that are enforced by social sanctions: a broader category than a social dilemma). Although she does not put it this way, hers is a generalization of the reputation account, allowing valuation of ongoing relationships to be egoistic or altruistic, and instrumental or intrinsic. An interdependent network of people expects to benefit from future interactions with one another. Within such a network, enforcement of a norm against harmful behavior reduces harm to people in the network, a direct benefit to each and, due to ongoing interdependence, each benefits indirectly from the reduction of harm to others. Norm enforcers are motivated because others in the network comply with a meta-norm to reward enforcers.
Her human-subject experiments nicely support her hypotheses, and provisional studies of natural settings are suggestive. Interdependence of actors does increase meta-norm enforcement, and thus compliance with the main norm. The stronger is interdependence, the stronger is the meta-norm of enforcement. Relational theory also predicts, and experimental tests illustrate, that meta-norm compliance arising from interdependence can lead to costly over-enforcement (as in political correctness), or to enforcement of norms harmful to the involved population, neither of which can be explained by consequentialist theories. Norms such as fashion are often explained by shared-meanings or typicality theories, but Horne argues that they are insufficient to explain the meta-norm of enforcement of such norms, and illustrates with experiment that relational theory can explain the meta-norm. And there is a wealth of related findings.
The book cleanly and concisely pursues the relational theory of norm enforcement. The prose is clear and simple (technical details are relegated to the appendices). The author carefully limits the scope of her hypotheses, and does not oversell her empirical results.
Horne's central experiment involves a first stage, when one actor is randomly selected and given the option contributing points to all three others, and a second stage where each of the four may contribute points to any set of the others. The second stage represents the interdependence motivating the meta-norm of rewarding compliance and punishing non-compliance of the actor in the first stage. In the second stage, under a condition of high interdependence, an individual's contribution of points to another is doubled; under a condition of low interdependence it is not.
The only important criticism I have of Horne's theory is that there is some imprecision in her interdependence construct: “people care about how others think of them and treat them” (p. 127), interdependence is “the value that people place on their relationships and the goods that they can attain from them [...] love, money, respect, and so forth” (p. 23), people in distress seek assistance from relatives (p. 24), people may benefit from improvements in others’ welfare (p. 38), one wants to demonstrate that one is a good exchange partner (p. 38), humans are group-oriented creatures (p. 78), and finally “rewarding encourages a future return of rewards” (p. 48, similarly p. 68). There is too much going on here. Further, the imprecision obscures what I suspect is the essential mechanism, an intrinsic reciprocity motivation, as shown by Fehr and coworkers (mentioned by Horne, p. 36, she says she supplements that consequentialist tenet with the finding that reciprocation is stronger with stronger interdependence). Fehr's reciprocity motivation, I suggest, best explains the reciprocity exhibited in the second stage of Horne's experiments.
The relational theory is avowedly a partial theory, and thus the following comments are external, not internal. I think research on social practices will better progress if we carefully distinguish moral norms, social norms, social conventions, and so-called descriptive or imitative norms, as Cristina Bicchieri does in the first chapter of her Grammar of Society. Next, the reputation and relational accounts share the platoon model of norm existence and compliance. Compliance is motivated by expectations of surveillance by known others. But compliance with moral norms does not essentially depend on surveillance; and undetected violation of even a trivial social norm can make one cringe with shame. Is norm frequency greater in anonymous Boston or in the densely connected mountain village of Montegrano? US individual income tax compliance, invisible to one's social network, is far higher than any cost-benefit calculation of enforcement can account for, a finding that drives economists crazy; but in Montegrano only a fool would contribute to the public good. Does higher density of social relationships predict higher compliance with moral norms and beneficial social norms? In the World Values Survey, stronger allegiance to family or religion predicts weaker trust of strangers (and lower income, education, health, etc.). Platoon models cannot account for the social practices ubiquitous within anonymous populations.
