Samid et al 2009 esa innsbruck talkPresentation Transcript
Costless Messages Outperform Altruistic Punishment Yuval Samid*, Ramzi Suleiman † , Tali Rainfeld-Rachlevsky † & Amir Hayek * Independent Researcher, 16 Shazar St., Haifa, Israel. † University of Haifa, Department of Psychology, Mount Carmel, Haifa, Israel. ‡ MCE-Systems, 51 Hamasger St, Tel-Aviv, Israel. We thank Keren Guz, Maayan Armelin, Fadi Farraj, Dina Baransi, Tamar Karanik and Maital Erlichman for their help in running the experiments Paper presented at the ESA Regional Meeting Europe 2009 , Innsbruck, Austria, SEPTEMBER 17-20
Studies on human cooperation show that, unlike other creatures, cooperation is frequent among genetically unrelated strangers, in non-repeated interactions and when gains from reputation are small or absent. Cooperation in these situations can be explained by altruistic punishment (Fehr & Gächter, 2002).
Previous research have demonstrated that:
Individuals engage in altruistic punishment, although it is costly.
Negative emotions towards defectors are the proximate mechanism behind altruistic punishment.
Altruistic punishment evolves even in large groups engaged in one-time, anonymous interactions.
Cooperation flourishes if altruistic punishment is possible, and breaks down if it is ruled out. Source: Fehr & G ä chter, 2002
Punishment has some serious drawbacks: 1) It is costly to punishers, hence it creates a second-order social dilemma (Yamagishi, 1987; Hauert et al., Science, 2007). 2) It bears no educational value. When punishment is terminated, the achieved cooperation is extinguished (Estes, 1944). 3) It undermines trust and confidence in voluntary cooperation (Mulder et. al., 2006)
In view of the disadvantages of punishment, we replaced the option of punishing other group members with the option of sending them costless, freely written messages.
Possible advantages of messages over punishment :
Since messages are costless, they can yield higher profits than the payoffs obtained by punishment.
While punishment can only serve as negative reinforcement, messages can be used as both positive and negative reinforcements.
Messages can bear an additional informative value, which might be beneficial for organizing cooperation.
Messages have an educational value, therefore their long-term effect on cooperation may be more pronounced than punishment.
Messages will be frequently used and will be addressed to both defectors ( negative verbal reinforcement ) and cooperators ( positive verbal reinforcement ).
The use of messages will enhance cooperation, or at least maintain it at its initial level.
When the option for sending messages is ruled out, cooperation will not brake down (educational effect).
Sending messages could yield higher profits than punishment.
First we replicated Fehr & Gächter’s (2002) Altruistic Punishment experiment. But punishment was not as effective. The weak punishment effect is attributed to cultural differences in punishing behavior (Herrmann, Thöni, Gächter, Science 2008, )
The effect of cultural differences in punishing behavior on cooperation in a public goods dilemma (Herrmann, Thöni, Gächter, Science , 2008) . For the punishment condition at Haifa, the average contribution was 9.3 (MU) , which puts us between Istanbul and Muscat .
Without punishment the cultural differences are less evident.
This cultural differences can be seen in the amount of punishment allocated to cooperators (also called Anti-social punishment ) .
In comparison with Zurich (Fehr & Gächter, 2002) , punishment expenditure was significantly higher at Haifa, 2.95 points per trial at Haifa , compared to 2.07 at Zurich ( Z = -2.51, P = 0.006, two-tailed Wilcoxon Two-Sample Test).
When comparing the allocation of punishment between free-riders and cooperators, we observe that higher levels of anti-social punishment were allotted in Haifa ( χ 2 (1) = 3.59, P = 0.0582, GEE Analysis) .
This cultural differences in Anti-social punishment
Expenditure on punishment as a function of the deviation of the contribution of the punished group member from the mean contribution of the other group members. Negative deviations = Free-riders – both slopes and difference between slopes are significant Positive deviations = Cooperators – difference between slopes is significant
Punishing behavior: Mean punishment expenditures for a given deviation from the punisher’s contribution.
Results of the message condition
With messages Without messages Mean contribution in the Message and Control condition Without messages With messages
Comparison of punishment , messages and control conditions
Comparison of the net profits in all conditions at Haifa and Zurich
Frequencies of messages and punishment utilization.
Examples of “anti-social” messages
Well done! But in the end you will loose a lot of money
Ooopss, try to be smart not righteous
You even don’t know who I am. Start living for yourself, and not for others
15 is agreed by all (sent to participants who contributed 20)
Optimism is nice!
1. Messages will be frequently used and will be addressed to both defectors and cooperators. Hypothesis: 2. The use of messages will enhance cooperation, or at least maintain it at its initial level. 4. Even if less effective than punishment, sending messages could yield higher profits than punishment. 3. When the option for sending messages is ruled out, cooperation will not brake down (educational effect). Confirmed Confirmed Rejected Confirmed
Summary and Main Conclusions
Punishment in Haifa was not effective.
2. Verbal messages in Haifa outperformed altruistic punishment. 4. A Combination of punishment and messages could prove cost-efficient. 3. It remains to investigate whether verbal messages outperform altruistic punishment in cultures with documented punishment efficacy.
Thank you for listening
מידת הענישה האנטי - סוציאלית ( ענישת תורמים ) יכולה להסביר את חוסר ההשפעה של ענישה בחברות שונות ( Herrmann, Thöni, Gächter, 2008 Science )