Team reasoning-Experimental evidence on cooperation from centipede games.

Johann Graf Lambsdorff,Katharina Werner,Marcus Giamattei,Manuel Schubert,Valerio Capraro

doi:10.1371/journal.pone.0206666

Abstract

Previous laboratory studies on the centipede game have found that subjects exhibit surprisingly high levels of cooperation. Across disciplines, it has recently been highlighted that these high levels of cooperation might be explained by “team reasoning”, the willingness to think as a team rather than as an individual. We run an experiment with a standard centipede game as a baseline. In two treatments, we seek to induce team reasoning by making a joint goal salient. First, we implement a probabilistic variant of the centipede game that makes it easy to identify a joint goal. Second, we frame the game as a situation where a team of two soccer players attempts to score a goal. This frame increases the salience even more. Compared to the baseline, our treatments induce higher levels of cooperation. In a second experiment, we obtain similar evidence in a more natural environment–a beer garden during the 2014 FIFA Soccer World Cup. Our study contributes to understanding how a salient goal can support cooperation.

Highlights

Cooperation has fascinated scientists from many disciplines for a long time
For the sake of collective welfare, humans are frequently willing to incur personal costs even if they could unilaterally benefit from defection. This has often been related to altruism towards genetically related group members or to long-term gains that can be reaped in repeated interaction [1]
Laboratory experiments have shown that levels of cooperation are high even when people play anonymously with partners that are unknown and unrelated to them and even when interaction is one-shot such that cooperation promises no future benefits from repeated interaction [2, 3, 4]

Summary

Introduction

Cooperation has fascinated scientists from many disciplines for a long time. For the sake of collective welfare, humans are frequently willing to incur personal costs even if they could unilaterally benefit from defection. To the best of our knowledge, not been investigated yet is whether variations in the experimental design can make a difference even if they leave expected payoffs unaffected For this purpose, we ran a standard centipede game (baseline) and two further treatments where we increase the salience of a joint goal. Subjects play a probabilistic variant of the centipede game where both players are made aware of a joint goal (a joint payoff of €200 in our experiment), with the probability of reaching the goal increasing across stages. Like in “Probabilistic”, the successful scorer would earn €160 while the other player would receive only €40 This description as a game of soccer supports the salience of a joint goal even more. 2 subjects noted that they had different instructions than others. 16 subjects noticed similar instructions and 305 did not take notice of others’ instructions

Results and econometric approach

Limitations

Discussion