Search

The strong reciprocity model of human cooperation (SRM) argues that strong reciprocators, who cooperate with others and punish non-cooperators, sustain within-group cooperation. However, the assumption that altruism and punishment are products of the same psychological mechanism of strong reciprocity has not been fully verified. Second-party punishment, for example as measured through rejection of unfair offers in the ultimatum game, has been demonstrated to have no relationship with cooperation in the prisoner's dilemma and other games. In this study, we tested the assumption of the SRM by comparing the participants' levels of cooperation in the prisoner's dilemma game and their inclination for punishment in a third-party punishment game. Non-student recruited from the general population (N = 182) participated in the study. The results show a weak but positive correlation between cooperation and third-party punishment, which is consistent with the SRM model.

In addition to the cost of punishment, the fear that others would evaluate punishers negatively can be a major obstacle for resolving the second-order social dilemma or failure of providing sanctions useful for solving a social dilemma problem. In an experiment with 81 participants, we tested whether providing information that other participants were in favor of punishing non-cooperators in a social dilemma situation would enhance cooperation in the second-order dilemma (i.e., punishment of non-cooperators). Participants received feedback of three bogus "participants" choices in a four-person social dilemma, in which one bogus participant defected and two others cooperated, and then received a chance to punish the sole non-cooperator. The hypothesis was supported among those who were motivated to punish the non-cooperator. They punished the non-cooperator when they were informed that the other participants also wanted to punish the non-cooperator. The feedback information that the other participants wanted to punish the non-cooperator induced the participants who were not motivated to punish the non-cooperator to punish less.

Cooperation in interdependent relationships is based on reciprocity in repeated interactions. However, cooperation in one-shot relationships cannot be explained by reciprocity. Frank, Gilovich, & Regan (1993) argued that cooperative behavior in one-shot interactions can be adaptive if cooperators displayed particular signals and people were able to distinguish cooperators from non-cooperators by decoding these signals. We argue that attractiveness and facial expressiveness are signals of cooperators. We conducted an experiment to examine if these signals influence the detection accuracy of cooperative behavior. Our participants (blind to the target's behavior in a Trust Game) viewed 30-seconds video-clips. Each video-clip was comprised of a cooperator and a non-cooperator in a Trust Game. The participants judged which one of the pair gave more money to the other participant. We found that participants were able to detect cooperators with a higher accuracy than chance. Furthermore, participants rated male non-cooperators as more attractive than male cooperators, and rated cooperators more expressive than non-cooperators. Further analyses showed that attractiveness inhibited detection accuracy while facial expressiveness fostered it.

A one-shot sequential prisoner's dilemma game with an in-group and an out-group member was conducted to test the group heuristic hypothesis for the in-group bias in minimal groups. Eighty-nine participants played the role of a second player and faced a fully cooperative first player. The results showed that in-group bias occurred only in the common knowledge condition in which the first player and the second player mutually knew each other's group membership, but not in the private knowledge condition in which the first player did not know the second player's group membership. These results provide support for the group heuristic hypothesis that in-group bias is an adaptive strategy in an assumed generalized exchange system to avoid a bad reputation as a defector.

An ultimatum game and two impunity games, in which rejection by the Responder had no impact on the Proposer's earnings, were conducted with 228 participants. The impunity game was run in two conditions: with feedback information, where the Responder's choice was disclosed to the Proposer, and without feedback information, where the Responder's choice was not disclosed to the Proposer. These two conditions were used to determine whether the motivation for rejection in the impunity game was social punishment or identity protection. The three game conditions and the Responder or Proposer roles were manipulated between participants. The rejection rates in the two impunity games were about half the rate in the ultimatum game: there was no difference in the rejection rates between the two feedback conditions of the impunity game. These results indicate that rejection in the ultimatum game is largely based on identity protection. The adaptive role of maintaining self-image as a commitment device was discussed.

The purpose of this study is to show that expectations of generalized reciprocity within one's own group are responsible for in-group trust. To test this hypothesis, an allocator choice game in the minimal group situation was used. We assigned the role of a recipient in a dictator game to all 81 subjects, and measured whether they chose to be a recipient of either an in-group or an out-group "allocator" who freely allocate a fixed reward between him/her and a recipient. The results indicate that in-group trust occurs only in the condition in which recipients know that allocators make a reward allocation knowing the group membership of their recipient; recipients show no preference for either an in-group or an out-group allocator when allocators make the decision without knowing the group membership of their recipient. It is thus shown that participants' in-group trust is derived from the general belief that people treat in-group members more favorably than out-group members-a belief about generalized reciprocity within groups.

Yamagishi and colleague showed that behavior favoring the in-group and trust of in-group members are based on the expectation of generalized exchange. We successfully tested a related prediction that people show stronger trust of in-group members than of out-group members in the faith game but not in the trust game. We argued that the direct exchange aspect inherent in the trust game suppresses expectations of generalized exchange taking place in the group. We conducted an experiment with 79 Japanese and 83 Australian participants. In the trust game, 27 of the 82 participants trusted the in-group trustee, while 29 trusted the out-group trustee. In the faith game, in which the expectation of generalized exchange was intact, 47 of the 80 participants trusted the in-group trustee and 37 trusted the out-group trustee. These results support our argument that expectations of generalized exchange, not in-group stereotypes, are the source of in-group trust. No cultural differences were observed between Japanese and Australian participants.

American and Japanese students, 44 and 38, respectively, participated in an experiment, and played a game together in seven- or eight-person groups. The game was a repeated version of bilateral trust game: Prisoner's Dilemma (PD) with choice of dependence. In no-information condition, participants were not told that some of the group members were from another country. In information condition, they were told that about half of the members were Japanese and the rest were Americans. We examined whether or not people trusted ingroup members (those from the same country) more than outgroup members, and whether or not they cooperated with ingroup members more than outgroup members. We found no evidence of ingroup bias in terms of trust and cooperation, and we did not find significant differences in the levels of trust or cooperation between those found among Americans and those among Japanese. On the other hand, it was found that American participants were more sensitive than Japanese counterparts, to information regarding the past trust behavior of other players when they were deciding whom they trust and whether or not they reciprocate another's trust.

Seventy-six participants first played a prisoner's dilemna (PD) game with an anonymous partner and then predicted behavior of other participants in the same PD game. In addition, they took a test of accuracy in judging sociometric relations among group members. Of the four factors of empathy, the two reflecting emotional aspects--self-directed and other-directed emotional responses--were found to have no relation with accuracy in predicting behavior of other participants and judging sociometric relations among group members. In contrast, the other two reflecting cognitive aspects--imagination and perspective taking--were found to be related to accuracy in predicting behavior of other participants in the PD game. Of the two, imagination was a better indicator of accuracy in the prediction than perspective taking. Implications of these findings are discussed in terms of domain specificities of adaptive task that different aspects of empathy are thought to fulfill.

Results of three experiments, in which a total of 167 students participated, consistently indicated that participants performed recognition tasks better for face photographs of defectors than those of cooperators. The face photographs used in the experiments were those of participants taken during prisoner's dilemma (PD) experiments a few years prior to the present study. In Experiments 1 and 2, photographs of cooperators and defectors in a one-shot PD experiment, taken after they had filled out a lengthy post-experimental questionnaire, were used. In Experiment 3, the photographs were those of high and low cooperators, taken at the moment of a cooperation or defection choice, respectively. Recognition was better for photographs that were judged unattractive than attractive. At the same time, it was better for photographs of less cooperative participants in the PD studies than those of more cooperative participants. Implications of the findings for Cosmides & Tooby's (1992) 'cheater-detection' module for social exchange domain were discussed.

Two alternative explanations for ingroup favoritism in the minimal group situation have been proposed: social identity and expectations of bounded generalized reciprocity. In this study, predictions derived from the two were examined for ingroup favoritism in a realistic group salutation with ongoing interactions. Results of the experiment, a replication of Karp, Jin, Yamagishi, and Shinotsuka (1993) which used groups with ongoing interactions rather than minimal groups, were inconsistent with either of the two explanations. (1) Ingroup favoritism emerged even in the condition under which the nature of fate control was unilateral. The finding was in sharp contrast to results of previous minimal groups experiments; the studies found ingroup favoritism only when fate control was mutual. (2) Analysis of post-experimental questionnaire responses indicated that participants engaged in ingroup favoritism, not to maximize differences between ingroup and outgroup but to improve ingroup members' gain. (3) Analysis of participants' open-ended self-reports revealed that they engaged in ingroup favoring reward allocation because of their personal bonds and likings for ingroup members as individuals rather than of their identification with the group per se.

This study examined the relationship between cooperation and trust in interpersonal trust formation. Previous studies of trust and cooperation using prisoner's dilemma (PD) games failed, methodologically as well as conceptually, to distinguish the former from the latter. In response to the criticism on the use of iterated PD games, and for the purpose of investigating dynamic relationship between trust and cooperation, an improvement in research methodology was recently proposed: namely, PD with choice of dependence (PD/D; Kakiuchi & Yamagishi, 1997; Yamagishi & Kakiuchi, 2000). We conducted an experiment to compare formation of trust relations in PD and PD/D. Result indicated a higher level of cooperation in PD/D than in PD. Further analysis of strategies used to build trustful relationship, where two partners trusted each other and reciprocated the other's trusting behavior, revealed that participants in PD/D adopted "cautious and unconditional cooperation strategy" rather than TFT strategy.

Recent computer simulation studies (Takahashi & Yamagishi, 1995, 1996) suggest the possibility that altruists outperform egoists, and the key to their advantage is a strategy called Downward OFT. The strategy calls for acting always altruistically, at the same time selecting as the target of the altruism the most altruistic person among potential recipients. The main purpose of this study was to examine whether people actually follow Downward OFT in a situation where no direct reciprocity was possible. Results were as follows: (1) Pure generalized exchange (i.e., unilateral recourse giving) did emerge and persisted over forty-five trials. (2) Subjects selected other altruists as recipients of their unilateral resource giving. (3) The tendency of altruistic subjects to discriminate among potential recipients of their altruism was stronger than that of self-interested subjects. (4) Subjects who unilaterally gave more resource ended up with more profit (resource received minus resource given) than those who gave less.

Contrary to the common sense idea that trustful people are gullible and easily believe whatever other people may say, past research reviewed by Rotter (1980a) indicated the idea was not necessarily valid. Two experiments of this paper further demonstrated that trustful people are more sensitive to information that indicates lack of trustworthiness in other people. In the experiments, subjects read a series of stories in which a person was about to make a choice between a trustworthy action and an untrustworthy one. Some of them were also given pieces of information regarding trustworthiness of the person. They were then asked to predict the likelihood of the person taking a trustworthy action. When no information was given, high trusters predicted with higher probability than low trusters that the person would take a trustworthy action. On the other hand, the high trusters lowered the predicted likelihood more steeply than the low when negative information was provided.

Experimental gaming researchers have recently realized that a larger context is important in which dyadic prisoner's dilemma (PD) relations are embedded. The main purpose of this study was to examine whether people facing a "selective-play" situation spontaneously adopt the out-for-tat strategy (OFT), which previous studies found the most effective in such a situation. OFT proscribes to (1) always cooperate, (2) keep playing with the current player as long as he/she cooperates, and (3) desert the partner as soon as he/she stops cooperating. Results of an experiment with 90 students, of groups of six or eight, showed that subjects who faced a high level of social uncertainty in fact spontaneously adopted the strategy, interacting with the same partner as long as he/she cooperated, and exiting from the relation when the partner defected. Subjects who faced a low level of social uncertainty kept only the third part of the strategy, leaving the relation with a defector, but not necessarily keeping interaction with a cooperative partner.

Two experiments examined the effect of illusion of control on in-group favoritism found in the minimal group situation (Tajfel, Billig, Bundy, & Flament, 1971). In bilateral dependency condition, each member made allocation decisions for in-group as well as out-group participants. It was exactly the same situation used in the original studies under the minimal group paradigm, and the subjects knew that their reward allocation too depended on others' decisions. In contrast, in unilateral dependency condition, the subjects made allocation decisions knowing that theirs were not dependent on others' decisions. In Experiment 1, an in-group bias in reward distribution was found in the bilateral dependency condition, but not in the unilateral condition. In Experiment 2, it was found that only those who felt illusion of control exhibited such an in-group bias. Results of the experiments therefore confirmed that illusion of control explained in-group favoritism, as Karp, Jin, Yamagishi, and Shinotsuka (1993) originally hypothesized.

Subjects performed a task in pairs and received false feedback concerning their performance levels. The performance level of one subject (HS) was set twice as high as the other (LS). It was predicted that (1) when the total rewards to be distributed between the two are dependent upon the sum of their performance scores, HS's and LS's equity judgments would agree with each other, but (2) when the total rewards are predetermined and not dependent upon their performance scores, HS would think that he/she deserves more than half of the rewards while LS would think that he/she deserves about half of the rewards. It was further predicted that the compromise between the two concerning reward distribution would be easier when accepting the partner's claim still yields referentially fair rewards to the subject than not. These predictions were supported in an experiment using 144 subjects in 72 same-sex pairs.