7. Five interesting facts about experimental economics games
We will now sketch a framework for interpreting experimental economics games results based on the following five facts. These facts highlight the important distinction between the explicit and implicit features of experimental games. By ‘explicit’ we mean the oral and written instructions to the players about the game, including any assurances of anonymity. By ‘implicit’ we mean all other information that might potentially influence play.
7.1. Explicit features
First, and most importantly, explicit features of games, such as opportunities for reputation building and punishment, can dramatically influence game play in ways predicted by rational choice theory (e.g., Fehr and Gächter, 2000 E. Fehr and S. Gächter, Cooperation and punishment in public goods experiments, Amer. Econom. Rev. 90 (2000), pp. 980–994. View Record in Scopus | Cited By in Scopus (237)Fehr and Gächter, 2000). This is strong evidence that the structure of the games is not completely opaque to the players.
7.2. Implicit features
Second, relatively subtle cues that are independent of the formal structure of an experimental game can nonetheless have a significant impact on players’ contributions. Kurzban (2001), for example, found that in a public goods game, brief, oblique eye contact or light taps on a shoulder or arm before each round significantly increased contributions by males (but not females), sometimes by more than 70%. He speculates that in ancestral environments these behaviors might have served as psychophysical cues of cooperation that would have been particularly important for men in hunting and warfare. In another example of the importance of implicit features, Haley and Fessler (2005) found that in a dictator game played on computer workstations, players using workstations with stylized eye-spots as part of the desktop background image were 55% more generous than players using workstations whose desktop background image did not contain eyespots. Haley and Fessler conjecture that eyespots enhanced players’ (perhaps unconscious) perception that their decisions would have reputational consequences. These examples raise the possibility that players in experimental economics games are using a wide variety of implicit cues that are not controlled by the experimenter to determine the extent to which they should cooperate in the game. If so, some of the altruism found in anonymous, one-shot games could be caused by such cues.
7.3. The importance of culture
Third, there is increasing evidence that culture plays an important role in the explanation of game results. The most intriguing finding from cross-cultural studies is that game play in many societies seems to reflect local social and economic institutions (Henrich et al., 2005). In the New Guinean societies of Au and Gnau, for example, accepting a gift creates a strong obligation to reciprocate, often in ways that the receiver finds onerous. If the receiver fails to reciprocate, he finds himself in a subordinate social position. Large gifts are consequently often refused. Perhaps not coincidentally, in these societies and unlike many other societies, large, ‘hyperfair’ offers exceeding 50% were frequently rejected in the ultimatum game (Tracer, 2003). This again illustrates that players draw upon information that is not specified by the formal structure of the game or the experimenter’s instructions.
7.4. Emotions
Fourth, rightly or wrongly, emotions (‘hot’ cognition) are often seen as fundamentally different from the rationality assumed to underly economic decision-making (‘cold’ cognition). Yet hot cognition plays a central role in the decisions of players in economics games, especially when it comes to punishing free-riders. In a public goods game with a punishment option, for example, the pattern of emotional responses to free riding was consistent with the hypothesis that strong negative emotions trigger the willingness to punish (Fehr and Gächter, 2000 and Fehr and Gächter, 2002).
7.5. Variation in game play
Finally, we are struck by the substantial individual variation in game play even within cultures. Across a number of different Western societies, there are large fractions of both free-riding and strongly reciprocating players (Fehr et al., 2002). Fischbacher et al. (2001), for example, found that in a public goods game in Switzerland, about 50% of players could be classified as conditionally cooperative with a self-serving bias; about 30% as purely selfish; and about 14% as conditionally cooperative when others contributed smaller amounts, but increasingly selfish when others contributed larger amounts. Kurzban and Houser (2005) similarly found in a public goods game in the US that 63% of players could be classified as reciprocators, 20% as free-riders, and 13% as cooperators (i.e., consistently generous).