- Email: [email protected]
Commitment bias: mistaken partner selection or ancient wisdom?
Evolution and Human Behavior 31 (2010) 22 – 28
Commitment bias: mistaken partner selection or ancient wisdom? István H. Back Interuniversity Center for Social Science Theory and Methodology, University of Groningen, The Netherlands Initial receipt 30 July 2008; final revision received 18 June 2009
Abstract Evidence across the social and behavioral sciences points to psychological mechanisms that facilitate the formation and maintenance of interpersonal commitment. In addition, evolutionary simulation studies suggest that a tendency for increased, seemingly irrational commitment is an important trait of successful exchange strategies. However, empirical research that tests corresponding psychological mechanisms is still largely lacking. Here an experimental test is proposed for one such mechanism, termed the commitment bias, which is hypothesized to increase people's commitment to existing partners beyond instrumental reasons. To exclude one alternative explanation, the commitment bias is distinguished from uncertainty reduction. Results from a cross-culturally replicated laboratory experiment (USA, China, and the Netherlands) provide support for the argument but also point to the importance of culture as an alternative or mediating factor. © 2010 Elsevier Inc. All rights reserved. Keywords: Commitment; Exchange; Interpersonal relationships; Uncertainty; Cross-cultural
1. Introduction Building and maintaining long-term social relationships is a natural part of human life and a behavior observed universally in all societies (Baumeister & Leary, 1995). People create relationships with great ease and are often reluctant to dissolve them, even if other alternatives are available. What makes people hold on to relationships in situations when this is apparently not in their best interest? According to standard rational choice explanations of long-term relationships, people repeatedly interact with each other when the benefits of a relationship outweigh its maintenance and alternative costs (Axelrod, 1984; Cook & Emerson, 1978; Hayashi & Yamagishi, 1998; Lawler & Yoon, 1996; Rusbult, Martz, & Agnew, 1998). First, having a long-term relationship with the same partner provides direct knowledge about the past behavior and thus the trustworthiness of a partner (“shadow of the past”). This makes a previous partner relatively more attractive than an unknown individual. Second, the prospect of recurring benefits from future mutual cooperation creates an incentive for existing partners to cooperate in the present, knowing that noncooperation could trigger retaliation by the partner and
E-mail address: [email protected]. 1090-5138/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.evolhumbehav.2009.06.006
thus jeopardize future payoffs (“shadow of the future,” see Axelrod, 1984). On the other hand, a great wealth of empirical evidence (cf. Baumeister & Leary, 1995) suggests that people have a tendency to become emotionally attached to each other, with less regard for the direct benefits derived from the relationship, its maintenance, or alternative costs. People create social relationships with great ease and strongly resist the dissolution of these relationships, well beyond rational considerations of practical advantage. A series of recent simulation studies explore the possibility of the evolution of a “commitment bias,” which could be an important psychological step in the process of commitment formation (Back & Flache, 2006, 2007, 2008; de Vos, Smaniotto, & Elsas, 2001; Smaniotto, 2004). These computational models are built on minimalistic assumptions about conditions of the ancestral environment, such as relatively small group size, food shortage, and increased need for mutual help. What these models show consistently is that when compared with strongly selfish or conditionally cooperative strategies, commitment emerges as the most viable strategy under a wide range of environmental conditions. For the rest of this article, “commitment” will refer to the behavior of repeatedly selecting the same partner even in the presence of potentially better alternative partners, and “commitment bias” will refer to a seemingly irrational
I.H. Back / Evolution and Human Behavior 31 (2010) 22–28
psychological tendency that is hypothesized to partially induce people to do so. The purpose of this article is to experimentally test the existence of a commitment bias. To do so, a short, anonymous, instrumental market setting is used where instrumental aspects of relationships are more salient, and at the same time the influence of non-instrumental factors can be more closely controlled. In such a setting, we have no reason to expect people to diverge from standard rational behavior toward commitment formation, unless there exists a systematic decision-making bias that pushes them to do so. 2. Hypotheses To test the existence of the commitment bias, I propose to test whether commitment behavior can be observed and systematically influenced under such an instrumental setting by manipulating only noninstrumental factors. A key factor that relevant for a commitment bias but not for instrumental rationality is the length of exposure between persons. In a world of instrumentally rational actors, simply because Person A has seen or interacted with Person B for some time should not, by itself, makes A more committed to B. In other words, an instrumentally rational A who assumes that everyone else obeys only their own self-interest would show equal preference for two different interaction partners B and C with whom interaction balances are equal. This is because A has no more incentive to trust B with whom he had a mutually cooperative interaction of $20, than another with whom he had two interactions of $10 each. In an ancestral world, however, the fact that someone has been around for a long time could have been a strong indicator for the trustworthiness, viability, and also the predictability of this person. Therefore, an actor with such an ancestral past should have a preference for a partner of two past interactions over a partner of only one interaction. Therefore: Hypothesis 1. (commitment bias) The longer the initial exposure between two persons, the more committed they become to each other, holding instrumental aspects of their relationship constant. A potential criticism against this argument is that during repeated interactions, uncertainty is reduced about the trustworthiness of the interaction partner. Reduction in uncertainty makes the long-term interaction partner always more attractive than an unknown stranger, holding other characteristics equal (see Kollock, 1994). Yamagishi, Cook, and Watabe (1998) proposed that uncertainty promotes commitment formation, and explained the tendency to form committed relationship with the individual's low level of general trust in others. The researchers showed in a crosscultural setting (comparing the USA and Japan) that those
23
who trust others less, in general, are more likely to form committed relationships. If the commitment bias is a purely psychological mechanism facilitating interpersonal commitment, it should exist independently from uncertainty reduction. Since it is impossible to fully eliminate uncertainty from an economic setting, I propose: Hypothesis 2. (uncertainty) The length of initial exposure has a positive effect on commitment even when controlling for the effect of uncertainty. If we indeed acquired a commitment bias during our evolutionary past, then this bias should be present in different cultures: Hypothesis 3. (cross-cultural) The commitment bias exists cross-culturally.
3. Design To test the hypotheses, I created a commitment-dilemma game that participants played during a laboratory experiment. To examine the existence of the commitment bias (Hypothesis 1) the experimental design manipulated the length of initial exposure between the participant and a steady partner at three levels (short, medium and long), and measured how committed the participants became to their partner. In order to test the relationship between uncertainty and commitment (Hypothesis 2), perceived uncertainty was measured using a post-experimental questionnaire. To test Hypothesis 3 about cross-cultural stability, I repeated the experiment in six locations of three countries—the Netherlands (Groningen and Utrecht), China (Shanghai and Nanjing), and the USA (Ithaca and Binghamton, NY, USA). The commitment game was played in a purely instrumental setting (buyer-seller relationships), involving only anonymous interactions. In addition to being more targeted, this setting also gives additional experimental control to exclude alternative explanations for commitment behavior, such as physical attraction, social control, or social desirability. During the commitment game, participants had to find trading partners. There were two kinds of actors in the game: artists who sold paintings and collectors who bought them. Participants were told that they would be randomly assigned to the role of seller (artist) or buyer (collector), but in reality, all participants were assigned to the role of seller, and played against computer-simulated buyers (see the original instructions in the Appendix). This ensured complete experimental control over the pricing mechanism and instrumental factors. Special care was taken to ensure that participants would not feel deceived during the experiment. Before starting the actual game, for example, participants had to wait for other
24
I.H. Back / Evolution and Human Behavior 31 (2010) 22–28
participants to reach the same phase in the game; then again between each round there was a short period of synchronization; incoming offers arrived with stochastically determined time durations, as if they were made by real people. To test for the effectiveness of the deception, we asked participants in a non-suggestive question after the game to guess the proportion of all participants who were acting as buyers in the laboratory at the moment. Those who were suspicious about the deception said 0%, or explicitly specified their doubt. We also asked participants to give their opinion about the behavior of other players whom they interacted with, and whether they thought other players behaved unusually in any way. Replies showed that less than 10% of participants had doubts that they were playing against real opponents present in the laboratory. After each experimental session, participants were debriefed in detail about the purpose and methodology of the experiment, including the fact of the deception. The game consisted of 17 consecutive rounds, and in each round the participant had to sell a new painting to one of four different buyers who made different bids on it (see a screen shot of the game in the electronic supplemental material). To create a commitment dilemma, those buyers who were not sold a painting by the participant were replaced by new buyers in the next round and did not appear again for the rest of the game in an identifiable fashion. Since there were always less than 17×3+1 participants in the lab, participants were told that those partners who once disappeared might reappear in a later round but would have a different name which would make it impossible for either party to recognize the other. The commitment dilemma manifested itself when the participant had been selling to an old buyer for a number of rounds and a newcomer made a better offer. The dilemma was between choosing the old partner with a suboptimal offer and getting a higher offer from a stranger, thereby also sacrificing the existing relationship with the old partner. Before such a commitment dilemma occurred, the length of initial exposure of the participant to the steady interaction partner had to be manipulated. Manipulating interpersonal relationships in the laboratory is not without a challenge. In fact, many previous experimenters dealing with commitment downright avoided doing so by explicitly assigning partners to each other or excluding the possibility of partner selection. To manipulate initial exposure in the current experiment, one of the computer-simulated buyers were programmed to give best offers successively for a number of rounds, before another buyer would appear with a better offer. At this very initial stage of the game most of the participants were expected to simply choose the best available option, which kept coming from the same partner. The actual length of the manipulation was one, two, or three best offers (rounds). The appearance of the steady partner with repeated best offers was preceded by three rounds of non-repeated best offers (i.e., best offers always came from different partners). Out of all participants, only 3% did not choose the best offer during
the manipulation of initial exposure (including the preceding 3 rounds), and these were excluded from the data set. The manipulation was designed such that the monetary benefit resulting from interaction with the steady partner was equal across conditions. In other words, no matter for how many rounds the same steady partner kept giving the best offer, the overall benefit of staying was equal across conditions, i.e., the cumulative difference between the best and second best alternative across the initial exposure (the opportunity cost of not choosing the best offer) was equal. If the participant remained committed in a commitment dilemma (did not choose a newcomer with the best offer), the steady interaction partner started alternating between providing best and second best offers in successive rounds. The price gap between best and second best offers continuously widened, resulting in an ever increasing sacrifice of profit for a committed participant. The main dependent variable, commitment, was simply measured as the number of interactions the participant kept interacting with the first steady interaction partner before deciding to switch to a new partner. In addition to measuring commitment, participants were also asked to report how positive, friendly or annoyed they felt towards each buyer in critical rounds that contained a commitment dilemma. The measurement of these emotional dispositions took place after participants received offers from buyers but before they had to make their choice. The measurement was administered identically across experimental conditions, and there is no reason to assume that it influenced the dependent variable differently in different conditions. After the game the level of perceived uncertainty was measured on a five-point scale by asking “How certain did you feel that you could sell your painting at a reasonable price?”. 3.1. Procedure The experiment was administered in computer laboratories with each participant sitting behind a computer, separated from others by panels or cubicles. Instructions (see Appendix), except for a brief initial welcome note by the experimenter, were received on screen, and all responses were given using a keyboard and mouse. The program was originally created in English and translated into Dutch and Mandarin by native speakers. The translation was then reviewed by separate native speakers for translation errors. To ensure the equivalence of the Mandarin translation, it was reverse-translated into English and double-checked against the original English text for any inconsistencies. Participants were recruited using colorful posters, advertisements on the faculty Web site, and in-class announcements. Participants were paid in local currency after the experiment, receiving a fixed amount for showing-up and a variable amount based on the size of virtual money gathered during the game. The amount of payoff was adjusted in each country to equal on average the price of three meals in the local university canteen.
I.H. Back / Evolution and Human Behavior 31 (2010) 22–28
3.2. Sample
25
Table 2 Regression Model 2 (N=309, R2=0.068)
The sample consisted of 326 participants from the three countries, the Netherlands (73 participants), China (206 participants) and the USA (47 participants). Of participants, 61.4% were female and 94.5% undergraduate students, mostly in humanity majors; 9.8% of all participants were suspicious about the deception (in the Netherlands 26%, in China only 4%, in the USA 13%). After a practice game but before playing for real stakes, participants had to complete a comprehension test about the rules of the game and the use of the game interface. If they incorrectly answered a question, they received feedback and explanation. Altogether, 10 people (9 in China, 1 in the USA) who did not choose the best price during the manipulation of the initial exposure were excluded from the analysis. 4. Results An immediately apparent finding from the experiment is that many participants avoided maximizing their profit through simple egoistic play. Although on average, participants chose more often the profitable alternative instead of the commitment alternative, 26.5% of all participants chose more often to be committed in situations when their old partner did not offer the best price. Moreover, although the game was fully anonymous, and the number of the total and current round was displayed on the participants' screen at all times during the game, 28.1% of the participants chose to be committed even in the last round of the game (“end-game violators”), when any future benefit from reciprocation, had to be ruled out (note, however, cross-cultural differences below.) 4.1. Hypotheses 1 and 2—commitment bias and uncertainty Hypothesis 1 prescribes a positive relationship between initial exposure and commitment. Initial exposure was manipulated at three levels (one, two, or three best offers), while commitment was measured by the number of rounds the participant stayed with the initial partner, who initially made the best offer(s). As expected by the hypothesis, the longer the initial exposure was, the more committed the participant became [r(314)=.143, p=.011]. To control for the effect of uncertainty, a linear regression model was fitted with commitment as a dependent variable, and initial exposure and perceived uncertainty (measured on a five-point scale) as explanatory variables. Regression
D.V.: Commitment
(Constant) Initial exposure Perceived uncertainty Exposure×uncertainty
Unstd. Coefficients B
S.E.
6.916 0.670 −0.933 0.539
0.575 0.292 0.284 0.239
t
Sig.
12.030 2.299 −3.283 2.253
.000 .022 .001 .025
Model 1 (see Table 1) shows a positive effect of the length of initial exposure on commitment behavior, when controlling for the effect of perceived uncertainty. This means that the longer people were initially exposed to each other, the more committed they became, controlling for the effect of uncertainty and instrumental benefits. This finding clearly supports Hypothesis 1. It is also clear from these results that, contrary to previous work (e.g., Kollock, 1994), perceived uncertainty decreased commitment. As it turned out from participants' verbal explanations (see later “Self-reported reasons for exit”), instead of staying committed to one partner, many participants protected themselves against uncertainty by trying to diversify exchange partners. Note that this occurred despite the fact that participants were not able to return to previous interaction partners once they switched. In addition, a positive interaction effect was found between initial exposure and perceived uncertainty (see Model 2 in Table 2). It turns out that although uncertainty decreased commitment, the strength of commitment between different levels of uncertainty converged, and was even reversed in the case of long initial exposure and little uncertainty (see Fig. 1). In other words, although uncertainty in itself decreased commitment, the commitment-inducing effect of a long relationship was strongest under high uncertainty.
Table 1 Regression Model 1 (N=310, R2=0.053) D.V.: Commitment
(Constant) Initial exposure Perceived uncertainty
Unstd. coefficients B
S.E.
6.983 0.655 −0.981
0.578 0.293 0.285
D.V. = dependent variable; Unstd. = unstandardized.
t
Sig.
12.084 2.233 −3.438
.000 .026 .001
Fig. 1. Interaction between initial exposure and perceived uncertainty (0=very certain, 4=very uncertain).
26
I.H. Back / Evolution and Human Behavior 31 (2010) 22–28
Table 3 Regression Model 3 (N=308, R2=0.095) D.V.: Commitment
Unstd. Coefficients
(Constant) Initial exposure Perceived uncertainty Exposure×uncertainty Chinese
B
S.E.
5.818 0.776 −0.890 0.480 1.487
0.676 0.290 0.281 0.237 0.497
t
Sig.
8.610 2.677 −3.167 2.023 2.993
.000 .008 .002 .044 .003
4.2. Hypothesis 3—cross-cultural stability To test whether the commitment bias exists crossculturally, the association between initial exposure and commitment had to be examined in the separate subsamples of each country. I found that the longer the initial exposure was, the more committed participants became in China [r (195)=.189, pb.001] and in the USA [r(44)=.365, p=.013], but no significant relationship was found in the Netherlands. At the same time, the average level of commitment was substantially higher in China [M=6.731, S.D.=4.8015], than in the Netherlands [M=5.274, S.D.=3.5208] and in the USA (M=5.217, S.D.=3.5208). The proportion of end-game violators—those who remained committed in the very last round, violating the rationality principle of noncooperation at the end of the game —was also considerably higher in China (37%) than in the Netherlands (11%) and the USA (11%). In addition, the Chinese were almost three times as likely (17.8%) to stay with their initial interaction partner throughout the entire game than the Americans (6.5%), with the Dutch closer to the Americans (6.8%). After including a dummy variable for China in the regression analysis (Model 3), I found a large positive effect of being Chinese on commitment, even when controlling for exposure, uncertainty and their interaction effect (Table 3). Apparent differences were found across cultures when participants had to describe their own behavior and preferences during the game. In a postexperimental questionnaire, participants were asked to rate on three separate five-point scales how important it was for them to play in a committed, fair, and profit-maximizing way. While the Chinese claimed that commitment, fairness but also profit was roughly equally important for them during the game, the US and the Dutch participants said that especially profit, but not fairness or commitment, was important (see Table 4).
Fig. 2. Reasons to leave steady partner.
In conclusion, the experiment uncovered both similarities and differences across cultures in commitment behavior, and found indication for the commitment bias in two out of three cultures. To obtain a more conclusive result, the crosscultural sample would need to be significantly extended. 4.3. Self-reported reasons for exit After participants exited from their relationship with the initial steady interaction partner, they were asked in an openended question, “Why did you choose to sell your painting to [the new buyer]?”. The majority (62%) of answers referred in some way to getting a higher profit. Next to these, however, a considerable proportion indicated a preference to build a network of buyers (e.g., “I will not sell to only one buyer, this way I gain an expanded influence through a little loss”, “make more friends and do more business”). Note that this was impossible in a strict sense, since once an interaction partner was abandoned, it would not return for the rest of the game under the same name. Other participants seem to have fallen victim to the fundamental attribution error (Jones and Harris 1967) and made bold inferences about buyers' personality from the price offered (e.g., “he offered the most money, and seems like a nice person to me” or “the offer is higher than the cost, which shows sincerity, the highest offer among all the buyers”). It also happened among some Chinese participants that they belittled their own performance and used this as an explanation for switching partners (e.g., “I am too shy to let
Table 4 Means of self-reported preferences across countries
China Netherlands USA Total
Committed (M)
N
S.D.
Profit (M)
N
S.D.
Fair (M)
N
S.D.
3.13 1.89 1.98 2.68
195 71 46 312
0.89 1.15 1.16 1.15
3.10 3.42 3.52 3.23
197 71 46 314
0.61 0.55 0.55 0.61
3.07 1.82 2.02 2.63
197 72 46 315
0.69 0.97 1.06 1.00
I.H. Back / Evolution and Human Behavior 31 (2010) 22–28
buyer 8 buy my paintings at the highest price”, or “I don't think this painting is very valuable”). A classification of the reasons to abandon the steady partner is shown on Fig. 2.
5. Discussion and conclusions A substantial body of evidence suggests that humans have a cognitive-emotional framework that facilitates creating and keeping interpersonal relationships, beyond rational considerations (Back & Flache, 2006, 2007, 2008; Baumeister & Leary, 1995; de Vos et al., 2001; Pedersen, 2004; Silk, 2003). However, empirical research targeted at uncovering related proximate mechanisms and, in particular, separating these mechanisms from competing explanations such as standard rational choice and uncertainty reduction are still largely lacking. This work aims to contribute by testing one such mechanism, that is termed here as the commitment bias. The present results suggest that people, in at least two out of the three countries examined, have a preference for staying committed to previous interaction partners and that past or future benefits resulting from such relationships are not sufficient to explain this behavior. This is indicated by the positive effect of initial exposure on the subsequent level of commitment even when controlling for uncertainty reduction (through regression analysis) and materialistic benefits (through the experimental design). The commitment bias may reflect a key step in the psychological process of commitment and long-term relationship formation. Of particular significance to this notion is the relatively large proportion of participants, who sacrificed profit by staying committed to their steady partner even in the final round of the game, showing what can only be termed as an irrational attachment under the circumstances. Results also suggest that although the level of uncertainty has a negative effect on commitment (i.e., people avoid uncertainty by searching through alternatives), the positive effect of a long-term relationship is strongest under high uncertainty. This is partially in contradiction with rational choice literature (Kollock, 1994, Yamagishi et al., 1998) and suggests that the relationship between uncertainty and commitment is more complex than described there. A possible explanation is that different types of uncertainties, i.e., uncertainty about the trustworthiness of partners (social uncertainty) and uncertainty about opportunities, have different effects on networking strategies. More specifically, it seems plausible that whereas social uncertainty leads people to become committed to few, trustworthy partners, uncertainty about opportunities leads people to explore and repeatedly interact with many partners (see more on this in Back, 2007). To examine a possible evolutionary argument for the commitment bias, I collected cross-cultural data in three countries. Results indicate that the commitment bias exists
27
in at least two of the three countries examined, China, and the USA, favoring the evolutionary argument. The facts that no significant effect is found in the Netherlands, and that mean commitment and “end-game commitment” are both much higher in China than in the other two counties, however, suggest that a purely evolutionary account may miss some of the picture. Henrich, Boyd, Bowles, Fehr, & Gintis, (2004), for example, show that cultural differences between groups may have substantial effects on exchange behavior, in addition to the effect of underlying evolved psychological mechanisms. Cross-cultural studies about relationships show that Asians tend to think of themselves as a member of a group, and in terms of their relationships with other people (“the interdependent self”), whereas Americans tend to think of themselves as individuals (“the independent self”) (see Markus & Kitayama, 1991). This may explain some of the differences found with regard to commitment between the samples from different countries. 5.1. Limitations and future research One shortcoming of the current experimental design is that uncertainty was measured, instead of experimentally manipulated. The drawback is that only a correlational relationship can be established between uncertainty and commitment and so the direction of causality cannot be ultimately decided. It may well be, for example, that people showing a certain level of commitment acted in response to perceived uncertainty, but also that they perceived a certain level of uncertainty as a consequence of being committed. Even so, the finding that certain participants had a strong preference for building a network of interaction partners— even though this was not possible within the framework of the experiment—clearly contradicts the findings of Kollock and (Yamagishi et al., 1998) about the positive effect of uncertainty on commitment. Therefore, an attractive path for further research is to examine more closely the relationship between commitment and uncertainty. In particular, prior research has explored the link between commitment and social uncertainty (Kollock, 1994; Yamagishi et al., 1998). Anthropological works show that people in several societies construct their social environment to adapt to uncertainties of the natural environment (see Wiessner, 1982, 2002). Whereas previous experimental work explored the link between social uncertainty and commitment, to my knowledge, there is no experimental work that systematically relates commitment to resource uncertainty or the inability of interaction partners to provide desired outcomes. With regard to the current cross-cultural argument about the commitment bias, further data collected from populations outside of the usual university pools would likely help to either strengthen the evolutionary argument, or to offer new insights about the cultural variability of commitment in exchange.
28
I.H. Back / Evolution and Human Behavior 31 (2010) 22–28
Acknowledgments I would like to thank Rita Smaniotto for her help and ideas throughout the planning and implementation of the experiments, Jurre van den Berg (University of Groningen) and Vincent Buskens (Utrecht University) for their assistance in conducting the experiments in the Netherlands; Xu Longshun (Fudan University), Zhao Kanglian (Nanjing University), Fan Xuejuan (East China Normal University), and Ji Wenxi for their help in running experiments in China; Michael Macy (Cornell University) and David Sloan Wilson (Binghamton University) for their insightful comments on the results and their help in conducting experiments in the USA. Further thanks go to participants of the Siena/Vidi discussion group at the University of Groningen; and finally Andreas Flache, Tom Snijders and Henk de Vos. This research was made possible by an Ubbo Emmius grant from the University of Groningen, and a Reisbeurs grant from the Netherlands Scientific Organization (NWO). Appendix A. Supplementary data Supplementary data associated with this article can be found in the online version, at doi:10.1016/j.evolhumbehav. 2009.06.006. References Axelrod, R. (1984). The Evolution of Cooperation. New York: Basic Books. Back, I. (2007). Commitment and evolution: Connecting emotion and reason in long-term relationships. PhD thesis, University of Groningen. Back, I., & Flache, A. (2006). The viability of cooperation based on interpersonal commitment. Journal of Artificial Societies and Social Simulation, 9(1), 12. Back, I., & Flache, A. (2007). Commitment, fairness and inequality. In C. Hernández, K. G. Troitzsch, & B. Edmonds (Eds.), Social Simulation: Technologies, advances and new discoveries : Idea Group Inc.
Back, I., & Flache, A. (2008). The adaptive rationality of interpersonal commitment. Rationality and Society. Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117(3), 497−529. Cook, K. S., & Emerson, R. M. (1978). Power, equity and commitment in exchange networks. American Sociological Review, 43, 721−739. de Vos, H., Smaniotto, R., & Elsas, D. A. (2001). Reciprocal altruism under conditions of partner selection. Rationality and Society, 13(2), 139−183. Hayashi, N., & Yamagishi, T. (1998). Selective play: Choosing partners in an uncertain world. Personality and Social Psychology Review, 2, 276−289. Henrich, J., Boyd, R., Bowles, S., Fehr, E., & Gintis, H. (Eds.). (2004). Foundations of human sociality: Economic experiments and ethnographic evidence in 15 small-scale societies. Oxford: Oxford University Press. Kollock, P. (1994). The emergence of exchange structures: An experimental study of uncertainty, commitment, and trust. American Journal of Sociology, 100(2), 313−345. Lawler, E., & Yoon, J. (1996). Commitment in exchange relations: Test of a theory of relational cohesion. American Sociological Review, 61(1), 89−108. Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychologal Review, 98(April), 224−253. Pedersen, C. A. (2004). How love evolved from sex and gave birth to intelligence and human nature. Journal of Bioeconomics, 6, 39−63. Rusbult, C., Martz, J., & Agnew, C. (1998). The investment model scale: measuring commitment level, satisfaction level, quality of alternatives, and investment size. Personal Relationships, 5(4), 357−391. Silk, J. (2003). Cooperation without counting: The puzzle of friendship. In P. Hammerstein (Ed.), Genetic and cultural evolution of cooperation (pp. 37−54). Cambridge, MA: MIT Press. Smaniotto, R.C. (2004). ‘You scratch my back and I scratch yours’ versus ‘love thy neighbour’: Two proximate mechanisms of reciprocal altruism. PhD thesis, ICS/University of Groningen. Wiessner, P. (1982). Risk, reciprocity, and social influences on! Kung San economics. In E. Leacock, & R. B. Lee (Eds.), Politics and history in band societies (pp. 61−84). Cambridge: Cambridge University Press. Wiessner, P. (2002). Hunting, healing, and hxaro exchange: A long-term perspective on ! Kung (Ju/'hoansi) large-game hunting. Evolution and Human Behavior, 23(6), 407−436. Yamagishi, T., Cook, K. S., & Watabe, M. (1998). Uncertainty, trust, and commitment formation in the United States and Japan. American Journal of Sociology, 104(1), 165−194.
Commitment bias: mistaken partner selection or ancient wisdom? István H. Back Interuniversity Center for Social Science Theory and Methodology, University of Groningen, The Netherlands Initial receipt 30 July 2008; final revision received 18 June 2009
Abstract Evidence across the social and behavioral sciences points to psychological mechanisms that facilitate the formation and maintenance of interpersonal commitment. In addition, evolutionary simulation studies suggest that a tendency for increased, seemingly irrational commitment is an important trait of successful exchange strategies. However, empirical research that tests corresponding psychological mechanisms is still largely lacking. Here an experimental test is proposed for one such mechanism, termed the commitment bias, which is hypothesized to increase people's commitment to existing partners beyond instrumental reasons. To exclude one alternative explanation, the commitment bias is distinguished from uncertainty reduction. Results from a cross-culturally replicated laboratory experiment (USA, China, and the Netherlands) provide support for the argument but also point to the importance of culture as an alternative or mediating factor. © 2010 Elsevier Inc. All rights reserved. Keywords: Commitment; Exchange; Interpersonal relationships; Uncertainty; Cross-cultural
1. Introduction Building and maintaining long-term social relationships is a natural part of human life and a behavior observed universally in all societies (Baumeister & Leary, 1995). People create relationships with great ease and are often reluctant to dissolve them, even if other alternatives are available. What makes people hold on to relationships in situations when this is apparently not in their best interest? According to standard rational choice explanations of long-term relationships, people repeatedly interact with each other when the benefits of a relationship outweigh its maintenance and alternative costs (Axelrod, 1984; Cook & Emerson, 1978; Hayashi & Yamagishi, 1998; Lawler & Yoon, 1996; Rusbult, Martz, & Agnew, 1998). First, having a long-term relationship with the same partner provides direct knowledge about the past behavior and thus the trustworthiness of a partner (“shadow of the past”). This makes a previous partner relatively more attractive than an unknown individual. Second, the prospect of recurring benefits from future mutual cooperation creates an incentive for existing partners to cooperate in the present, knowing that noncooperation could trigger retaliation by the partner and
E-mail address: [email protected]. 1090-5138/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.evolhumbehav.2009.06.006
thus jeopardize future payoffs (“shadow of the future,” see Axelrod, 1984). On the other hand, a great wealth of empirical evidence (cf. Baumeister & Leary, 1995) suggests that people have a tendency to become emotionally attached to each other, with less regard for the direct benefits derived from the relationship, its maintenance, or alternative costs. People create social relationships with great ease and strongly resist the dissolution of these relationships, well beyond rational considerations of practical advantage. A series of recent simulation studies explore the possibility of the evolution of a “commitment bias,” which could be an important psychological step in the process of commitment formation (Back & Flache, 2006, 2007, 2008; de Vos, Smaniotto, & Elsas, 2001; Smaniotto, 2004). These computational models are built on minimalistic assumptions about conditions of the ancestral environment, such as relatively small group size, food shortage, and increased need for mutual help. What these models show consistently is that when compared with strongly selfish or conditionally cooperative strategies, commitment emerges as the most viable strategy under a wide range of environmental conditions. For the rest of this article, “commitment” will refer to the behavior of repeatedly selecting the same partner even in the presence of potentially better alternative partners, and “commitment bias” will refer to a seemingly irrational
I.H. Back / Evolution and Human Behavior 31 (2010) 22–28
psychological tendency that is hypothesized to partially induce people to do so. The purpose of this article is to experimentally test the existence of a commitment bias. To do so, a short, anonymous, instrumental market setting is used where instrumental aspects of relationships are more salient, and at the same time the influence of non-instrumental factors can be more closely controlled. In such a setting, we have no reason to expect people to diverge from standard rational behavior toward commitment formation, unless there exists a systematic decision-making bias that pushes them to do so. 2. Hypotheses To test the existence of the commitment bias, I propose to test whether commitment behavior can be observed and systematically influenced under such an instrumental setting by manipulating only noninstrumental factors. A key factor that relevant for a commitment bias but not for instrumental rationality is the length of exposure between persons. In a world of instrumentally rational actors, simply because Person A has seen or interacted with Person B for some time should not, by itself, makes A more committed to B. In other words, an instrumentally rational A who assumes that everyone else obeys only their own self-interest would show equal preference for two different interaction partners B and C with whom interaction balances are equal. This is because A has no more incentive to trust B with whom he had a mutually cooperative interaction of $20, than another with whom he had two interactions of $10 each. In an ancestral world, however, the fact that someone has been around for a long time could have been a strong indicator for the trustworthiness, viability, and also the predictability of this person. Therefore, an actor with such an ancestral past should have a preference for a partner of two past interactions over a partner of only one interaction. Therefore: Hypothesis 1. (commitment bias) The longer the initial exposure between two persons, the more committed they become to each other, holding instrumental aspects of their relationship constant. A potential criticism against this argument is that during repeated interactions, uncertainty is reduced about the trustworthiness of the interaction partner. Reduction in uncertainty makes the long-term interaction partner always more attractive than an unknown stranger, holding other characteristics equal (see Kollock, 1994). Yamagishi, Cook, and Watabe (1998) proposed that uncertainty promotes commitment formation, and explained the tendency to form committed relationship with the individual's low level of general trust in others. The researchers showed in a crosscultural setting (comparing the USA and Japan) that those
23
who trust others less, in general, are more likely to form committed relationships. If the commitment bias is a purely psychological mechanism facilitating interpersonal commitment, it should exist independently from uncertainty reduction. Since it is impossible to fully eliminate uncertainty from an economic setting, I propose: Hypothesis 2. (uncertainty) The length of initial exposure has a positive effect on commitment even when controlling for the effect of uncertainty. If we indeed acquired a commitment bias during our evolutionary past, then this bias should be present in different cultures: Hypothesis 3. (cross-cultural) The commitment bias exists cross-culturally.
3. Design To test the hypotheses, I created a commitment-dilemma game that participants played during a laboratory experiment. To examine the existence of the commitment bias (Hypothesis 1) the experimental design manipulated the length of initial exposure between the participant and a steady partner at three levels (short, medium and long), and measured how committed the participants became to their partner. In order to test the relationship between uncertainty and commitment (Hypothesis 2), perceived uncertainty was measured using a post-experimental questionnaire. To test Hypothesis 3 about cross-cultural stability, I repeated the experiment in six locations of three countries—the Netherlands (Groningen and Utrecht), China (Shanghai and Nanjing), and the USA (Ithaca and Binghamton, NY, USA). The commitment game was played in a purely instrumental setting (buyer-seller relationships), involving only anonymous interactions. In addition to being more targeted, this setting also gives additional experimental control to exclude alternative explanations for commitment behavior, such as physical attraction, social control, or social desirability. During the commitment game, participants had to find trading partners. There were two kinds of actors in the game: artists who sold paintings and collectors who bought them. Participants were told that they would be randomly assigned to the role of seller (artist) or buyer (collector), but in reality, all participants were assigned to the role of seller, and played against computer-simulated buyers (see the original instructions in the Appendix). This ensured complete experimental control over the pricing mechanism and instrumental factors. Special care was taken to ensure that participants would not feel deceived during the experiment. Before starting the actual game, for example, participants had to wait for other
24
I.H. Back / Evolution and Human Behavior 31 (2010) 22–28
participants to reach the same phase in the game; then again between each round there was a short period of synchronization; incoming offers arrived with stochastically determined time durations, as if they were made by real people. To test for the effectiveness of the deception, we asked participants in a non-suggestive question after the game to guess the proportion of all participants who were acting as buyers in the laboratory at the moment. Those who were suspicious about the deception said 0%, or explicitly specified their doubt. We also asked participants to give their opinion about the behavior of other players whom they interacted with, and whether they thought other players behaved unusually in any way. Replies showed that less than 10% of participants had doubts that they were playing against real opponents present in the laboratory. After each experimental session, participants were debriefed in detail about the purpose and methodology of the experiment, including the fact of the deception. The game consisted of 17 consecutive rounds, and in each round the participant had to sell a new painting to one of four different buyers who made different bids on it (see a screen shot of the game in the electronic supplemental material). To create a commitment dilemma, those buyers who were not sold a painting by the participant were replaced by new buyers in the next round and did not appear again for the rest of the game in an identifiable fashion. Since there were always less than 17×3+1 participants in the lab, participants were told that those partners who once disappeared might reappear in a later round but would have a different name which would make it impossible for either party to recognize the other. The commitment dilemma manifested itself when the participant had been selling to an old buyer for a number of rounds and a newcomer made a better offer. The dilemma was between choosing the old partner with a suboptimal offer and getting a higher offer from a stranger, thereby also sacrificing the existing relationship with the old partner. Before such a commitment dilemma occurred, the length of initial exposure of the participant to the steady interaction partner had to be manipulated. Manipulating interpersonal relationships in the laboratory is not without a challenge. In fact, many previous experimenters dealing with commitment downright avoided doing so by explicitly assigning partners to each other or excluding the possibility of partner selection. To manipulate initial exposure in the current experiment, one of the computer-simulated buyers were programmed to give best offers successively for a number of rounds, before another buyer would appear with a better offer. At this very initial stage of the game most of the participants were expected to simply choose the best available option, which kept coming from the same partner. The actual length of the manipulation was one, two, or three best offers (rounds). The appearance of the steady partner with repeated best offers was preceded by three rounds of non-repeated best offers (i.e., best offers always came from different partners). Out of all participants, only 3% did not choose the best offer during
the manipulation of initial exposure (including the preceding 3 rounds), and these were excluded from the data set. The manipulation was designed such that the monetary benefit resulting from interaction with the steady partner was equal across conditions. In other words, no matter for how many rounds the same steady partner kept giving the best offer, the overall benefit of staying was equal across conditions, i.e., the cumulative difference between the best and second best alternative across the initial exposure (the opportunity cost of not choosing the best offer) was equal. If the participant remained committed in a commitment dilemma (did not choose a newcomer with the best offer), the steady interaction partner started alternating between providing best and second best offers in successive rounds. The price gap between best and second best offers continuously widened, resulting in an ever increasing sacrifice of profit for a committed participant. The main dependent variable, commitment, was simply measured as the number of interactions the participant kept interacting with the first steady interaction partner before deciding to switch to a new partner. In addition to measuring commitment, participants were also asked to report how positive, friendly or annoyed they felt towards each buyer in critical rounds that contained a commitment dilemma. The measurement of these emotional dispositions took place after participants received offers from buyers but before they had to make their choice. The measurement was administered identically across experimental conditions, and there is no reason to assume that it influenced the dependent variable differently in different conditions. After the game the level of perceived uncertainty was measured on a five-point scale by asking “How certain did you feel that you could sell your painting at a reasonable price?”. 3.1. Procedure The experiment was administered in computer laboratories with each participant sitting behind a computer, separated from others by panels or cubicles. Instructions (see Appendix), except for a brief initial welcome note by the experimenter, were received on screen, and all responses were given using a keyboard and mouse. The program was originally created in English and translated into Dutch and Mandarin by native speakers. The translation was then reviewed by separate native speakers for translation errors. To ensure the equivalence of the Mandarin translation, it was reverse-translated into English and double-checked against the original English text for any inconsistencies. Participants were recruited using colorful posters, advertisements on the faculty Web site, and in-class announcements. Participants were paid in local currency after the experiment, receiving a fixed amount for showing-up and a variable amount based on the size of virtual money gathered during the game. The amount of payoff was adjusted in each country to equal on average the price of three meals in the local university canteen.
I.H. Back / Evolution and Human Behavior 31 (2010) 22–28
3.2. Sample
25
Table 2 Regression Model 2 (N=309, R2=0.068)
The sample consisted of 326 participants from the three countries, the Netherlands (73 participants), China (206 participants) and the USA (47 participants). Of participants, 61.4% were female and 94.5% undergraduate students, mostly in humanity majors; 9.8% of all participants were suspicious about the deception (in the Netherlands 26%, in China only 4%, in the USA 13%). After a practice game but before playing for real stakes, participants had to complete a comprehension test about the rules of the game and the use of the game interface. If they incorrectly answered a question, they received feedback and explanation. Altogether, 10 people (9 in China, 1 in the USA) who did not choose the best price during the manipulation of the initial exposure were excluded from the analysis. 4. Results An immediately apparent finding from the experiment is that many participants avoided maximizing their profit through simple egoistic play. Although on average, participants chose more often the profitable alternative instead of the commitment alternative, 26.5% of all participants chose more often to be committed in situations when their old partner did not offer the best price. Moreover, although the game was fully anonymous, and the number of the total and current round was displayed on the participants' screen at all times during the game, 28.1% of the participants chose to be committed even in the last round of the game (“end-game violators”), when any future benefit from reciprocation, had to be ruled out (note, however, cross-cultural differences below.) 4.1. Hypotheses 1 and 2—commitment bias and uncertainty Hypothesis 1 prescribes a positive relationship between initial exposure and commitment. Initial exposure was manipulated at three levels (one, two, or three best offers), while commitment was measured by the number of rounds the participant stayed with the initial partner, who initially made the best offer(s). As expected by the hypothesis, the longer the initial exposure was, the more committed the participant became [r(314)=.143, p=.011]. To control for the effect of uncertainty, a linear regression model was fitted with commitment as a dependent variable, and initial exposure and perceived uncertainty (measured on a five-point scale) as explanatory variables. Regression
D.V.: Commitment
(Constant) Initial exposure Perceived uncertainty Exposure×uncertainty
Unstd. Coefficients B
S.E.
6.916 0.670 −0.933 0.539
0.575 0.292 0.284 0.239
t
Sig.
12.030 2.299 −3.283 2.253
.000 .022 .001 .025
Model 1 (see Table 1) shows a positive effect of the length of initial exposure on commitment behavior, when controlling for the effect of perceived uncertainty. This means that the longer people were initially exposed to each other, the more committed they became, controlling for the effect of uncertainty and instrumental benefits. This finding clearly supports Hypothesis 1. It is also clear from these results that, contrary to previous work (e.g., Kollock, 1994), perceived uncertainty decreased commitment. As it turned out from participants' verbal explanations (see later “Self-reported reasons for exit”), instead of staying committed to one partner, many participants protected themselves against uncertainty by trying to diversify exchange partners. Note that this occurred despite the fact that participants were not able to return to previous interaction partners once they switched. In addition, a positive interaction effect was found between initial exposure and perceived uncertainty (see Model 2 in Table 2). It turns out that although uncertainty decreased commitment, the strength of commitment between different levels of uncertainty converged, and was even reversed in the case of long initial exposure and little uncertainty (see Fig. 1). In other words, although uncertainty in itself decreased commitment, the commitment-inducing effect of a long relationship was strongest under high uncertainty.
Table 1 Regression Model 1 (N=310, R2=0.053) D.V.: Commitment
(Constant) Initial exposure Perceived uncertainty
Unstd. coefficients B
S.E.
6.983 0.655 −0.981
0.578 0.293 0.285
D.V. = dependent variable; Unstd. = unstandardized.
t
Sig.
12.084 2.233 −3.438
.000 .026 .001
Fig. 1. Interaction between initial exposure and perceived uncertainty (0=very certain, 4=very uncertain).
26
I.H. Back / Evolution and Human Behavior 31 (2010) 22–28
Table 3 Regression Model 3 (N=308, R2=0.095) D.V.: Commitment
Unstd. Coefficients
(Constant) Initial exposure Perceived uncertainty Exposure×uncertainty Chinese
B
S.E.
5.818 0.776 −0.890 0.480 1.487
0.676 0.290 0.281 0.237 0.497
t
Sig.
8.610 2.677 −3.167 2.023 2.993
.000 .008 .002 .044 .003
4.2. Hypothesis 3—cross-cultural stability To test whether the commitment bias exists crossculturally, the association between initial exposure and commitment had to be examined in the separate subsamples of each country. I found that the longer the initial exposure was, the more committed participants became in China [r (195)=.189, pb.001] and in the USA [r(44)=.365, p=.013], but no significant relationship was found in the Netherlands. At the same time, the average level of commitment was substantially higher in China [M=6.731, S.D.=4.8015], than in the Netherlands [M=5.274, S.D.=3.5208] and in the USA (M=5.217, S.D.=3.5208). The proportion of end-game violators—those who remained committed in the very last round, violating the rationality principle of noncooperation at the end of the game —was also considerably higher in China (37%) than in the Netherlands (11%) and the USA (11%). In addition, the Chinese were almost three times as likely (17.8%) to stay with their initial interaction partner throughout the entire game than the Americans (6.5%), with the Dutch closer to the Americans (6.8%). After including a dummy variable for China in the regression analysis (Model 3), I found a large positive effect of being Chinese on commitment, even when controlling for exposure, uncertainty and their interaction effect (Table 3). Apparent differences were found across cultures when participants had to describe their own behavior and preferences during the game. In a postexperimental questionnaire, participants were asked to rate on three separate five-point scales how important it was for them to play in a committed, fair, and profit-maximizing way. While the Chinese claimed that commitment, fairness but also profit was roughly equally important for them during the game, the US and the Dutch participants said that especially profit, but not fairness or commitment, was important (see Table 4).
Fig. 2. Reasons to leave steady partner.
In conclusion, the experiment uncovered both similarities and differences across cultures in commitment behavior, and found indication for the commitment bias in two out of three cultures. To obtain a more conclusive result, the crosscultural sample would need to be significantly extended. 4.3. Self-reported reasons for exit After participants exited from their relationship with the initial steady interaction partner, they were asked in an openended question, “Why did you choose to sell your painting to [the new buyer]?”. The majority (62%) of answers referred in some way to getting a higher profit. Next to these, however, a considerable proportion indicated a preference to build a network of buyers (e.g., “I will not sell to only one buyer, this way I gain an expanded influence through a little loss”, “make more friends and do more business”). Note that this was impossible in a strict sense, since once an interaction partner was abandoned, it would not return for the rest of the game under the same name. Other participants seem to have fallen victim to the fundamental attribution error (Jones and Harris 1967) and made bold inferences about buyers' personality from the price offered (e.g., “he offered the most money, and seems like a nice person to me” or “the offer is higher than the cost, which shows sincerity, the highest offer among all the buyers”). It also happened among some Chinese participants that they belittled their own performance and used this as an explanation for switching partners (e.g., “I am too shy to let
Table 4 Means of self-reported preferences across countries
China Netherlands USA Total
Committed (M)
N
S.D.
Profit (M)
N
S.D.
Fair (M)
N
S.D.
3.13 1.89 1.98 2.68
195 71 46 312
0.89 1.15 1.16 1.15
3.10 3.42 3.52 3.23
197 71 46 314
0.61 0.55 0.55 0.61
3.07 1.82 2.02 2.63
197 72 46 315
0.69 0.97 1.06 1.00
I.H. Back / Evolution and Human Behavior 31 (2010) 22–28
buyer 8 buy my paintings at the highest price”, or “I don't think this painting is very valuable”). A classification of the reasons to abandon the steady partner is shown on Fig. 2.
5. Discussion and conclusions A substantial body of evidence suggests that humans have a cognitive-emotional framework that facilitates creating and keeping interpersonal relationships, beyond rational considerations (Back & Flache, 2006, 2007, 2008; Baumeister & Leary, 1995; de Vos et al., 2001; Pedersen, 2004; Silk, 2003). However, empirical research targeted at uncovering related proximate mechanisms and, in particular, separating these mechanisms from competing explanations such as standard rational choice and uncertainty reduction are still largely lacking. This work aims to contribute by testing one such mechanism, that is termed here as the commitment bias. The present results suggest that people, in at least two out of the three countries examined, have a preference for staying committed to previous interaction partners and that past or future benefits resulting from such relationships are not sufficient to explain this behavior. This is indicated by the positive effect of initial exposure on the subsequent level of commitment even when controlling for uncertainty reduction (through regression analysis) and materialistic benefits (through the experimental design). The commitment bias may reflect a key step in the psychological process of commitment and long-term relationship formation. Of particular significance to this notion is the relatively large proportion of participants, who sacrificed profit by staying committed to their steady partner even in the final round of the game, showing what can only be termed as an irrational attachment under the circumstances. Results also suggest that although the level of uncertainty has a negative effect on commitment (i.e., people avoid uncertainty by searching through alternatives), the positive effect of a long-term relationship is strongest under high uncertainty. This is partially in contradiction with rational choice literature (Kollock, 1994, Yamagishi et al., 1998) and suggests that the relationship between uncertainty and commitment is more complex than described there. A possible explanation is that different types of uncertainties, i.e., uncertainty about the trustworthiness of partners (social uncertainty) and uncertainty about opportunities, have different effects on networking strategies. More specifically, it seems plausible that whereas social uncertainty leads people to become committed to few, trustworthy partners, uncertainty about opportunities leads people to explore and repeatedly interact with many partners (see more on this in Back, 2007). To examine a possible evolutionary argument for the commitment bias, I collected cross-cultural data in three countries. Results indicate that the commitment bias exists
27
in at least two of the three countries examined, China, and the USA, favoring the evolutionary argument. The facts that no significant effect is found in the Netherlands, and that mean commitment and “end-game commitment” are both much higher in China than in the other two counties, however, suggest that a purely evolutionary account may miss some of the picture. Henrich, Boyd, Bowles, Fehr, & Gintis, (2004), for example, show that cultural differences between groups may have substantial effects on exchange behavior, in addition to the effect of underlying evolved psychological mechanisms. Cross-cultural studies about relationships show that Asians tend to think of themselves as a member of a group, and in terms of their relationships with other people (“the interdependent self”), whereas Americans tend to think of themselves as individuals (“the independent self”) (see Markus & Kitayama, 1991). This may explain some of the differences found with regard to commitment between the samples from different countries. 5.1. Limitations and future research One shortcoming of the current experimental design is that uncertainty was measured, instead of experimentally manipulated. The drawback is that only a correlational relationship can be established between uncertainty and commitment and so the direction of causality cannot be ultimately decided. It may well be, for example, that people showing a certain level of commitment acted in response to perceived uncertainty, but also that they perceived a certain level of uncertainty as a consequence of being committed. Even so, the finding that certain participants had a strong preference for building a network of interaction partners— even though this was not possible within the framework of the experiment—clearly contradicts the findings of Kollock and (Yamagishi et al., 1998) about the positive effect of uncertainty on commitment. Therefore, an attractive path for further research is to examine more closely the relationship between commitment and uncertainty. In particular, prior research has explored the link between commitment and social uncertainty (Kollock, 1994; Yamagishi et al., 1998). Anthropological works show that people in several societies construct their social environment to adapt to uncertainties of the natural environment (see Wiessner, 1982, 2002). Whereas previous experimental work explored the link between social uncertainty and commitment, to my knowledge, there is no experimental work that systematically relates commitment to resource uncertainty or the inability of interaction partners to provide desired outcomes. With regard to the current cross-cultural argument about the commitment bias, further data collected from populations outside of the usual university pools would likely help to either strengthen the evolutionary argument, or to offer new insights about the cultural variability of commitment in exchange.
28
I.H. Back / Evolution and Human Behavior 31 (2010) 22–28
Acknowledgments I would like to thank Rita Smaniotto for her help and ideas throughout the planning and implementation of the experiments, Jurre van den Berg (University of Groningen) and Vincent Buskens (Utrecht University) for their assistance in conducting the experiments in the Netherlands; Xu Longshun (Fudan University), Zhao Kanglian (Nanjing University), Fan Xuejuan (East China Normal University), and Ji Wenxi for their help in running experiments in China; Michael Macy (Cornell University) and David Sloan Wilson (Binghamton University) for their insightful comments on the results and their help in conducting experiments in the USA. Further thanks go to participants of the Siena/Vidi discussion group at the University of Groningen; and finally Andreas Flache, Tom Snijders and Henk de Vos. This research was made possible by an Ubbo Emmius grant from the University of Groningen, and a Reisbeurs grant from the Netherlands Scientific Organization (NWO). Appendix A. Supplementary data Supplementary data associated with this article can be found in the online version, at doi:10.1016/j.evolhumbehav. 2009.06.006. References Axelrod, R. (1984). The Evolution of Cooperation. New York: Basic Books. Back, I. (2007). Commitment and evolution: Connecting emotion and reason in long-term relationships. PhD thesis, University of Groningen. Back, I., & Flache, A. (2006). The viability of cooperation based on interpersonal commitment. Journal of Artificial Societies and Social Simulation, 9(1), 12. Back, I., & Flache, A. (2007). Commitment, fairness and inequality. In C. Hernández, K. G. Troitzsch, & B. Edmonds (Eds.), Social Simulation: Technologies, advances and new discoveries : Idea Group Inc.
Back, I., & Flache, A. (2008). The adaptive rationality of interpersonal commitment. Rationality and Society. Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117(3), 497−529. Cook, K. S., & Emerson, R. M. (1978). Power, equity and commitment in exchange networks. American Sociological Review, 43, 721−739. de Vos, H., Smaniotto, R., & Elsas, D. A. (2001). Reciprocal altruism under conditions of partner selection. Rationality and Society, 13(2), 139−183. Hayashi, N., & Yamagishi, T. (1998). Selective play: Choosing partners in an uncertain world. Personality and Social Psychology Review, 2, 276−289. Henrich, J., Boyd, R., Bowles, S., Fehr, E., & Gintis, H. (Eds.). (2004). Foundations of human sociality: Economic experiments and ethnographic evidence in 15 small-scale societies. Oxford: Oxford University Press. Kollock, P. (1994). The emergence of exchange structures: An experimental study of uncertainty, commitment, and trust. American Journal of Sociology, 100(2), 313−345. Lawler, E., & Yoon, J. (1996). Commitment in exchange relations: Test of a theory of relational cohesion. American Sociological Review, 61(1), 89−108. Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychologal Review, 98(April), 224−253. Pedersen, C. A. (2004). How love evolved from sex and gave birth to intelligence and human nature. Journal of Bioeconomics, 6, 39−63. Rusbult, C., Martz, J., & Agnew, C. (1998). The investment model scale: measuring commitment level, satisfaction level, quality of alternatives, and investment size. Personal Relationships, 5(4), 357−391. Silk, J. (2003). Cooperation without counting: The puzzle of friendship. In P. Hammerstein (Ed.), Genetic and cultural evolution of cooperation (pp. 37−54). Cambridge, MA: MIT Press. Smaniotto, R.C. (2004). ‘You scratch my back and I scratch yours’ versus ‘love thy neighbour’: Two proximate mechanisms of reciprocal altruism. PhD thesis, ICS/University of Groningen. Wiessner, P. (1982). Risk, reciprocity, and social influences on! Kung San economics. In E. Leacock, & R. B. Lee (Eds.), Politics and history in band societies (pp. 61−84). Cambridge: Cambridge University Press. Wiessner, P. (2002). Hunting, healing, and hxaro exchange: A long-term perspective on ! Kung (Ju/'hoansi) large-game hunting. Evolution and Human Behavior, 23(6), 407−436. Yamagishi, T., Cook, K. S., & Watabe, M. (1998). Uncertainty, trust, and commitment formation in the United States and Japan. American Journal of Sociology, 104(1), 165−194.