The hard problem of cooperation

Based on individual variation in cooperative inclinations, we define the ‘‘hard problem of cooperation’’ as that of achievinghigh levels of cooperation in a group of non-cooperative types. Can the hard problem be solved by institutions withmonitoring and sanctions? In a laboratory experiment we find that the answer is affirmative if the institution is imposed onthe group but negative if development of the institution is left to the group to vote on. In the experiment, participants weredivided into groups of either cooperative types or non-cooperative types depending on their behavior in a public goodsgame. In these homogeneous groups they repeatedly played a public goods game regulated by an institution thatincorporated several of the key properties identified by Ostrom: operational rules, monitoring, rewards, punishments, and(in one condition) change of rules. When change of rules was not possible and punishments were set to be high, groups ofboth types generally abided by operational rules demanding high contributions to the common good, and therebyachieved high levels of payoffs. Under less severe rules, both types of groups did worse but non-cooperative types didworst. Thus, non-cooperative groups profited the most from being governed by an institution demanding highcontributions and employing high punishments. Nevertheless, in a condition where change of rules through voting wasmade possible, development of the institution in this direction was more often voted down in groups of non-cooperativetypes. We discuss the relevance of the hard problem and fit our results into a bigger picture of institutional and individualdeterminants of cooperative behavior