Monopolizing Sanctioning Power under Noise Eliminates Perverse Punishment But Does Not Increase Cooperation

Abstract

We run several experiments which allow us to compare cooperation under perfect and imperfect information in a centralized and decentralized punishment regime. Under perfect and extremely noisy information, aggregate behavior does not differ between institutions. Under intermediate noise, punishment escalates in the decentralized peer-to-peer punishment regime which badly affects efficiency while sustaining cooperation for longer. Only decentralized punishment is often directed at cooperators (perverse punishment). We report several, sometimes subtle, differences in punishment behavior, and how contributions react.

Consensual punishment does not promote cooperation in the six-person prisoner's dilemma game with noisy public monitoring

“…For example, it has been argued that pool-punishment institutions, where free-riders are punished by a central authority that is maintained through voluntary contributions [ 45 ], have played an important role in maintaining cooperation at a large scale [ 46 , 47 , 48 ]. An important but understudied aspect in this strand of research is the procedure by which a central authority decides whom to punish, in particular, if cooperation cannot perfectly be observed [ 28 ]. Our study contributes to this strand of research by comparing the effectiveness of individual and collective decision rules for the implementation of punishment to maintain cooperation in noisy environments.…”

“…Grechenig et al [ 27 ] and Fischer et al [ 28 ] experimentally examine how noise in the display of contributions affects the extent to which cooperation can be maintained through punishment institutions with an IDR. Both studies consider linear PGGs with noise.…”

“…This renders noise more ambiguous than it is in our setting, in which a misrepresented contribution is always shown as defection and vice versa. Grechenig et al [ 27 ] and Fischer et al [ 28 ] find that subjects do not refrain from employing punishment in the presence of noise, such that two types of errors occur in prosocial punishment. First, some punishments are mistakenly directed at cooperators.…”

“…It has been shown that noise can lead to misguided punishment decisions and thus limit the effectiveness of an IDR to sustain cooperation [ 27 , 28 ]. For example, let us assume that co-workers with different expertise work on a common project.…”

“…However, as some will misperceive A’s and/or B’s contributions, B may be criticized for free-riding, and A may not be sufficiently reprimanded for shirking. These sanctions or their absence, respectively, can occur even though A’s and B’s colleagues know that they might not observe A’s and B’s true efforts [ 27 , 28 ]. As a consequence, B may be discouraged from putting in the same amount of effort again, and A may be encouraged to continue shirking.…”