The behavior of people in the Ultimatum Game (UG) has been analyzed in terms of a two-person interaction between a proposer and a responder. Yet there is a third person involved: the experimenter who donates the money and has the power to withdraw it if the responder does not accept the proposer’s offer. Since the third person has no name in the theoretical analyses of the UG, let us call him or her the donor. Whether or not the donor can be treated as a neutral observer, equivalent to a measurement instrument, seems to be particularly interesting in cross-cultural comparisons. When an experimenter walks into one of the 15 small-scale societies, he or she represents a technologically advanced tribe and is likely to stand out more than when in a university lab. Both the proposer (she) and the responder (he) know that the donor will record their choices, and they might not be indifferent to the impression their behavior has on the donor. In addition, the responder may realize that by accepting he might take money away from the donor, whereas if he rejects the offer, he will give money to the donor. This three-body perspective differs from the theoretical treatment of the donor as a neutral figure, whose only task is to explain the rules and record the behavior. . .
[T]here are good reasons to consider the possibility that the behavior of the proposer is not simply a function of his expectations about the responder, or of some stable social preferences, but is also targeted at the donor. The UG is supposedly played anonymously, a term that describes the relation between proposer and responder, whereas in fact there is no anonymity between the two and the donor. In Henrich et al.’s analysis, a choice between a selfish or altruistic offer is assumed to reflect the proposer’s social preferences or expectations concerning the responder. Yet, seen as a three-body game, his or her choice could reflect her goals and expectations concerning the donor as well. A proposer might want to appear generous instead of greedy in the eyes of the donor (who is not anonymous) rather than before the responder (who is anonymous). A proposer may be embarrassed if the donor sees her offer being rejected. . .
The same holds for the responder, who can expect that the donor knows what amount he accepted or refused. A responder may also be concerned with creating a reputation of being tough by rejecting a low offer, or seeks social approval by not showing anger or disappointment in public and accepting any offer. Since he can assume he will never find out who the proposer was, and vice versa, the primary target of reputation building appears to be the donor rather than the proposer. In this view, fairness or toughness are signals towards the donor as well as the partner, unlike in two-body analyses of the UG. . .
The same perspective can be applied to the Dictator Game. The fact that a proposer offers more than zero in the Dictator game has been taken as the demonstration of genuine rather than strategic altruism. Seen as a three-body game, this conclusion does not follow. If the proposer is concerned with her reputation, an anonymous player who cannot identify her consequently cannot promote her reputation, whereas the non-anonymous donor can. . .
The three-body view of the UG extends to explanatory attempts in terms of social analogs. Such an explanation was proposed for the Orma, who recognized a similarity between the Public Goods Game and the local contributions Orma households make when the community decides to pursue a common good, such as building a school. For the UG, no such analogy was proposed, and we would be curious to learn from the authors whether they were never observed, or else, what analogies have been made. A threebody view invites looking for analogies with a richer interactive structure: one party donating goods to a second one, while retaining the option to withdraw them if the second party’s division of the pie is rejected by a third party. In such cultural analogs, if they exist, the donations could be bribes, gifts, alms, or obligations, or something else. And when the money changes hands from the donor to the proposer, it can change its functional category, such as from a gift to an obligation – but only when considered from the three-body perspective. If people can map the UG into a common analogy, then the variance in the offers (rejections) should decrease, whereas the absolute offers and acceptance levels will still vary with the specific analogy.
How would the behavior be different if the donor provided other goods to the proposer than money? If heuristics for sharing depend on the goods – meat and honey are meticulously shared among the Ache, but goods purchased by money are not (Henrich et al. 2004) – then the observed behavior in the UG should also depend on the kind of pie, not solely on some abstract preferences for selfish or altruistic behavior. In fact, in Lamalera, packs of cigarettes rather than money were used in the UG, and the Lamalera ranked among the top “altruistic” societies. Cigarettes tend to be shared, and this may enhance the appearance of a preference for altruism.
Henrich et al. assume that cultural evolution shapes preferences, yet the alternative to this view is that evolution shapes decision heuristics instead. A tit-for-tat player follows a heuristic, not a preference for altruism or defection, except in the first move. The resulting behavior is based on an interactive strategy, not on preferences that are assumed to be stable like personality traits. We think that the connection between cultural evolution and behavioral economics might be better understood as the shaping of heuristics in the adaptive toolbox. Here, the interpretation of the UG as a three-body transaction provides a new twist to the question of the influence of the environment in which the heuristics of the players are adapted. They may react to the donor as well as to the other player.