EFTA02690244.pdf

Submitted to Biology Letters biology letters Fear of shame drives cooperation Journal: Biology Letters Manuscript ID: Draft Article Type: Research Date Submitted by the Author: 'l e Complete List of Authors: Jacquet, Jennifer; University of British Columbia, Department of Mathematics Hauert, Christoph; University of British Columbia, Department of Mathematics Traulsen, Arne; Max-Planck-Institute for Evolutionary Biology, Research Group for Evolutionary Theory Milinski, Manfred; Max-Planck-Institute for Evolutionary Biology, Department of Evolutionary Ecology Subject: Behaviour < BIOLOGY, Evolution < BIOLOGY Categories: Evolutionary Biology cooperation, honour, shame, public goods game, tragedy of the Keywords: commons 1:5C _ r:V.M/L P:1, it • pl., http://mc.manuscriptcentral.com/bl EFTA_R1_02030115 EFTA02690244 Page 1 of 9 Submitted to Biology Letters Jacques, Haunt, Traulsen, & Milinski 1 2 3 4 Fear of shame drives cooperation 5 6 Jennifer Jacquetl2., Christoph Hauert', Arne Traulsen3 and Manfred Milinski4 7 8 Department of Mathematics, University of British Columbia, 1984 Mathematics Road, 9 10 11 Vancouver, British Columbia, V6T 1Z2 Canada (where the research was carried out) 12 13 2Sea Around Us Project, Aquatic Ecosystem Research Laboratory, 2202 Main Mall, Vancouver, 14 15 British Columbia, V6T 1Z4 Canada 16 17 18 3Research Group for Evolutionary Theory, Max-Planck-Institute for Evolutionary Biology, 19 20 August-Thienemann-Strasse 2, D-24306 Ploen, Germany 21 22 4Department of Evolutionary Ecology, Max-Planck-Institute for Evolutionary Biology, August- 2 24 25 Thienemann-Strasse 2, D-24306 Ploen, Germany 26 27 *Author for correspondence ([email protected]) 28 29 30 31 32 Summary 33 34 Can shame lead to greater cooperation? We test this hypothesis with anonymous 6-player public 35 36 37 goods experiments, an experimental paradigm used to investigate problems related to overusing 38 39 common resources. We instructed the players that the 2 individuals who were least generous 40 41 after 10 rounds would be exposed to the group. As the natural antithesis, we also test the effects 42 43 44 of honour, exposing only the 2 players who were most generous. The nonmonetary effects of 45 46 shame and honour each led to approximately 50% higher donations to the public good as 47 48 compared to the control, demonstrating that both shame and honour can drive cooperation and 49 50 51 help alleviate the tragedy of the commons. 52 53 54 55 Keywords: cooperation, honour, shame, public goods game, tragedy of the commons 56 57 58 59 60 http://mc.ma nuscriptcentral.com/bI EFTA_R1_02030116 EFTA02690245  Submitted to Biology Letters Page 2 of 9 Jacques, Haunt, Traulsen, & Milinski 1 2 3 1. INTRODUCTION 4 5 6 Shame is induced when offenders are singled out for public scorn. Social emotions such as 7 8 empathy and shame likely featured prominently in the early evolution of pro-social behavior in 9 10 11 hominids [I]. Today, as modern democratic societies have moved away from involving the 12 13 public in punishment [2], it is tempting to think of shame as a vestigial phenomenon from 14 15 medieval times, when the accused were placed in the town pillory or emblazoned with a scarlet 16 17 18 letter. But as digital technology increasingly allows us to communicate and keep track of one 19 20 another, we sense a resurgence of shame and ask whether shame remains a potential enforcement 21 22 for social behavior. Here we test experimentally if the fear of shame is an incentive to cooperate. 23 24 25 Social dilemmas arise through the consumption of common resources, such as wild fish, 26 27 fossil fuels, or clean water, and translate into a tragedy of the commons, where group 28 29 cooperation is undermined by individual self-interest [3]. Public goods experiments capture the 30 31 32 tension between individual and group-interest, and usually confirm Hardin's pessimistic promise 33 34 that "freedom in the commons brings ruin to all" [4]. In a typical setup, players receive start-up 35 36 37 capital and can choose to donate some or none of it to a 'public goods' project; donations are 38 39 increased by a given factor and redistributed evenly among all players, irrespective of whether 40 41 they contributed. Maximum net benefit is achieved if all players donate, but individual players 42 43 44 earn most if they keep their capital and profit off the generosity of the other players. Usually 45 46 players inevitably exercise this 'rational' self-interest and cooperation quickly declines. 47 48 Public goods interactions also exemplify cooperation's intricacies. For instance, players 49 50 51 are willing to pay to punish uncooperative behavior [5]. Experiments that allow players to build 52 53 and benefit monetarily from reputation lead to increased cooperation [6-7]. In games that offer 54 55 players anonymity. uncooperative behavior is more prevalent [8] while the opposite is true of 56 57 58 59 60 http://mc.ma nuscriptcentral.com/bI EFTA_R1_02030117 EFTA02690246 Page 3 of 9 Submitted to Biology Letters Jacques, Hauen, Traulsen, & Milinski 1 2 3 games in which players know that each of their decisions will be linked to their real identities [9- 4 5 6 I 1]. Revealing the identities of all participants corresponds to full transparency but does not 7 8 allow us to discern the effects of shame and its antithesis, honour. If players know that only the 9 10 11 least or most cooperative individuals are to stand in front of their peers, will they cooperate more 12 13 as a group? 14 15 We designed this public goods experiment to isolate the effects of shame as well as 16 17 18 honour, with no monetary consequences to either experience. We hypothesized that the threat of 19 20 both shame and honour would lead to increased public contributions. 21 22 23 24 25 2. MATERIALS AND METHODS 26 27 We tested our predictions with 180 first-year University of British Columbia science students divided into 3 28 29 treatments, shame, honour, and control, consisting of 10 identical 6-player games each. To foster indelibility for 30 31 shame and honour, all 6 players came from the same class to ensure that the players were acquainted with each other 32 33 and were recruited within the first few weeks of the term to ensure that they would meet again repeatedly during the 34 35 term. 36 37 There was a single group of 6 players in the room at a time. Players were partitioned off from each other as 38 39 well as the experimenters, who stayed out of view for the duration of the actual experiment. Each player received a 40 41 starting account of CDN$12 and a randomly assigned unique pseudonym (obscure Greek gods). Players were 42 43 anonymitcd. both to the experimenters and other players but players in the honour and shame treatments wrote real 44 45 names inside an envelope labeled with their pseudonym, which was collected by the experimenter so the 2 least 46 47 generous players (or most generous in the honour treatment) could eventually be identified. The box with the 48 49 concealed names remained visible to all players at all times to protect their anonymity. All 6 players could see a 50 51 public screen on which instructions and the game were projected. Before the game, an experimenter read the 52 instructions, and demonstrated the choices and outcomes in example games using pseudonyms not appearing in the 53 54 experiment. 55 56 57 58 59 60 http://mc.manuscriptcentral.comfbl EFTA_R1_02030118 EFTA02690247  Submitted to Biology Letters Page 4 of 9 Jacques, Haunt, Traulsen, 1 2 3 Players chose whether to contribute $1 into a public pool or keep it in his/her private funds at each round 4 5 for 12 rounds. Without visual contact with the player, an experimenter passed a locked box into each cubicle, in 6 7 which every player placed his/her anonymized envelope (blank on the outside; pseudonym on the inside) containing 8 9 $0 or SI. Contributions were recorded on the public screen under each player's pseudonym. The group total and 10 11 player payout were displayed for each round, as was the aggregate total contribution for each player. 12 13 After round 10. the experimenter opened the envelopes labeled with the pseudonyms of the 2 players who 14 15 donated least overall in the shame treatments to reveal their real names (in the honour treatment it was the 2 players 16 17 who donated most). In the event of a tie, the experimenter chose 2 players by throwing a 6-sided die, with the 18 19 pseudonyms pre-determinedly linked to each number. Ties occurred in 5 of the shame games and 4 of the honour 20 21 games. Interestingly, ties only occurred only for the second least (or most) generous players but never for the least 22 23 (or most) generous players. The 2 least generous players went in front of the group and wrote their name on a board 24 25 under the phrase "I donated least", which was visible for the entire game (for honour, the phrase was "I donated 26 27 most" and the 2 most generous players went in front). The real names of these 2 players were also added to the 28 29 pseudonyms on the public screen. The remaining 4 envelopes with the names of the 4 players that retained their 30 31 anonymity were visibly destroyed and discarded in front of the group. In the control treatment all 6 players remained 32 33 anonymous. At the end of round 12. each player left with the money he/she kept during the game plus the profits 34 35 from the public pool. Note that the profits from the public pool were the same for every player and could therefore 36 37 be distributed without compromising the players' anonymity. The students were asked not to discuss the experiment 38 39 with anyone else. 40 41 42 43 3. RESULTS 44 45 46 In each treatment, initial cooperation in the public goods game declined as expected (paired t-test 47 48 between l g and 10th round, tis=10, t= 2.71, P=0.024; nh=10, t=4.61, P.1.001; /lel°, 1= 7.61, 49 50 P<0.0001; the 6-player group is the statistical unit; all probabilities are 2-tailed; Fig. 1). 51 52 53 Donations for the first 10 rounds in the shame treatment were significantly higher as compared to 54 55 the control (2-sample t-test, ns=10, tic=10, t= 2.24, P=0.038), as were contributions in the honour 56 57 treatment (2-sample t-test, tih=10, tk=10, t= 2.89. P4).010). Average group contributions were 58 59 60 4 http://m c.ma nuscriptcentral.com/bI EFTA_R1_02030119 EFTA02690248 Page 5 of 9 Submitted to Biology Letters Jacques, Haunt, Traulsen, & Milinski 1 2 3 53% higher in the shame treatment ($33.8 $13.6 stdev) and 48% higher in honour ($32.6 4 5 6 $6.6 stdev) than in the control ($22.1 $9.4 stdev; full cooperation is $60 in donations). 7 8 Our results show that a promise to single out free-riding individuals for public scrutiny 9 10 11 can lead to greater cooperation from the whole group. Even in this one-off experiment, people 12 13 were willing to pay not necessarily to avoid exposure, but to avoid shame. In fact, players in the 14 15 honour treatment did not fear exposure; they paid for it. In contrast to the honour treatment, 16 17 18 group cooperation in the shame treatment significantly declined following round 10 (paired 1-test 19 20 between 10th and It round, r= 3.67, P4.005), corroborating our finding that the threat of being 21 22 singled out as a free rider had been driving cooperation. 23 24 25 26 27 4. DISCUSSION 28 29 Cues of being watched enhance cooperation [12] and when humans lived in small groups, 30 31 32 it was easy to observe individual behavior. However, as human society grew. gossip, by way of 33 34 language, replaced direct observation as a vector for keeping track of human behavior [13-14]. 35 36 37 At this transition, shame and honour could have been at a premium -- when the chance of 38 39 witnessing behavior firsthand was then amplified by the possibility that it could be verbally 40 41 expressed to the community. 42 43 44 Shame is an uncomfortable phenomenon in part because it invites the public in on the 45 46 punishment. Today, there are also convincing philosophical objections to a legal system that 47 48 shames individuals on the grounds that such punishments undermine human dignity [2]. But the 49 50 51 absence of shaming by the state does not preclude the absence of shame altogether in society, 52 53 especially as social media increases the frequency, speed, and inclusiveness of communication. 54 55 The Internet increasingly creates a global town square where gossip travels fast and where shame 56 57 58 59 60 5 http://mc.ma nuscriptcentral.com/bI EFTA_R1_02030120 EFTA02690249  Submitted to Biology Letters Page 6 of 9 Jacques, Haunt, Traulsen, & Milinski 1 2 3 and honour might experience resurgence. As the same time, the Internet is also a tool for 4 5 6 tracking compliance and for transparency [e.g. 15]. Transparency also enhances cooperation [9- 7 8 I I] but can be costly to provide and its use can be limited. Transparency requires time 9 10 11 evaluating and determining a satisfactory performance. This becomes increasingly difficult in 12 13 our current era, where human attention, not information, is a scarce resource [16]. By singling 14 15 out only the least or most cooperative players, shame and honour are more parsimonious than 16 17 18 full transparency and rely on social norms as reference points. 19 20 In this experiment, the fear of shame as well as the promise of honour led to increased 21 22 cooperation from the entire group and might even help transform a crowd into a community. 23 24 25 The results illuminate a potential positive consequence in the unavoidable revival of the old 26 27 threat of shame: to encourage groups to cooperate and maintain resources that we all share. 28 29 30 31 32 ACKNOWLEDGEMENTS 33 34 We thank the students of UBC for their participation, NSERC for its financial support, Y. Kondo 35 36 37 for her statistical advice, and M. Bailey and L. Wardil for affable assistance with the 38 39 experiments. 40 41 42 43 44 REFERENCES 45 46 1 Boyd, R. & Richerson, P.J. 2009 Culture and the evolution of human cooperation. Phil. Trans. 47 48 R. Soc. B 364, 3281-3288. (doi:10.1098/rstb.2009.0134) 49 50 51 2 Nussbaum, M. 2006 Hidingfrom Humanity: Disgust, Shame, and the Law, Princeton, New 52 53 Jersey: Princeton University Press. 54 55 56 57 58 59 60 6 http://mc.ma nuscriptcentral.com/bI EFTA_R1_02030121 EFTA02690250 Page 7 of 9 Submitted to Biology Letters Jacques, Haunt, Traulsen, & Milinski 1 2 3 3 Ostrom, E. 1990 Governing the Commons: The Evolution ofInstitutions for Collective Action 4 5 6 Cambridge, United Kingdom: Cambridge University Press. 7 8 4 Hardin, G. 1968 The tragedy of the commons Science 162, 1243. 9 10 11 5 Water, S, Renner, E. & Sefton, M. 2008 The long-run benefits of punishment Science 322, 12 13 1510. (doi:10.1126/science.I 164744) 14 15 6 Milinski, M., Semmann, D. & Krambeck, H.J. 2002 Reputation helps solve the 'tragedy of the 16 17 18 commons'. Nature 415, 424-426.(doi:10.1038/415424a) 19 20 7 Sylwester, K. & Roberts, G. Cooperators benefit through reputation-based partner choice in 21 22 economic games Biol. Lett. 6, 659-662. (doi:10.1098/rsb1.2010.0209) 23 24 25 8 Semmann, D., Krambeck, H.J. & M. Milinski, M. 2004 Strategic investment in reputation. 26 27 Behavioral Ecology and Sociobiology 56, 248-252. (doi:10.1007/s00265-004-0782-9) 28 29 9 Water, S. & E. Fehr 1999 Collective action as a social exchange. J. Econ. Behavior and 30 31 32 Organization 39, 341-369. (doi:10.1016/S0167-2681(99)00045-1) 33 34 10 Rege, M. & Telle, K. 2004 The impact of social approval and framing on cooperation in 35 36 37 public good situations. J. Public Econ. 88, 1625 - 1644. (doi:10.1016/S0047-2727(03)00021-5) 38 39 11 Andreoni, J. & Petrie, R. 2004 Public goods experiments without confidentiality: a glimpse 40 41 into fund-raising. J. Public Econ. 88, 1605 — 1623. (doi:10.1016/S00472727(03)00040-9) 42 43 44 12 Bateson, M., Nettles, D.. & Roberts, G. 2006 Cues of being watched enhance cooperation in 45 46 real-world setting. Biol. Len. 2, 412-414. (doi: 10.1098/rsb1.2006.0509) 47 48 13 Dunbar, R. 1997 Grooming. Gossip and the Evolution ofLanguage, Cambridge, 49 50 51 Massachusetts, Harvard University Press. 52 53 54 55 56 57 58 59 60 7 http://mc.ma nuscriptcentral.com/bI EFTA_R1_02030122 EFTA02690251  Submitted to Biology Letters Page 8 of 9 Jacques, Haunt, Traulsen, & Milinski 1 2 3 14 Sommerfeld, R.D., Krambeck, H.J. & Milinski, M. 2007 Gossip as an alternative for direct 4 5 6 observation in games of indirect reciprocity Proc. Natl. Acad. ScL U.S.A. 104. 17435 — 17440. 7 8 (doi: 10.1073/pnas.0704598104) 9 10 11 15 Fung, A., Graham, M. & Weil, D. 2007 Full Disclosure: The Perils and Promise of 12 13 Transparency, Cambridge, Massachusetts, Cambridge University Press. 14 15 16 Wu, F & Huberman, B.A. 2007 Novelty and collective action Proc. Natl. Acad. Set U.S.A. 16 17 18 104, 17599 — 17601. (doi: 10.1073/pnas.0704916104) 19 20 21 22 Figure caption: Average group contributions for each treatment: shame, honour, control. In the 23 24 25 shame treatment, the 2 least generous players were exposed as free riders after round 10 while in 26 27 honour the 2 most generous were revealed as highest contributors to the group. In the control 28 29 treatment, all players retained anonymity over the 12 rounds, as did the non-exposed players in 30 31 32 shame and honour. 33 34 35 36 37 Short title: Shame and cooperation 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 S http://mc.manuscriptcentral.com/bl EFTA_R1_02030123 EFTA02690252 Page 9 of 9 Submitted to Biology Letters 1 2 3 4 5 6 6 7 8 9 5 10 • • mean group contribution ($) 11 S • exposure 12 4 ••• • 13 14 "••o--- -- 15 -a- ---c--- -1°. % 3 "... 16 °•- ---o ••• MI •••••0 17 18 2 19 20 - D- Control 21 1 -0- Honour 22 -0- Shame 23 24 0 25 26 2 3 4 5 6 7 8 9 10 II 12 27 28 round in public goods gams 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 http://mc.manuscriptcentral.comfbl EFTA_R1_02030124 EFTA02690253