Network reciprocity and cooperation in laboratory experiments
Jelena Grujic, Imperial College London

The mechanisms behind the promotion of cooperation are still not very well understood. One of the mechanism proposed to explain the promotion of cooperation is network reciprocity. We present a series of laboratory experiments with human subjects designed to test the emergence of cooperation when humans play a Multiplayer Iterated Prisoner's Dilemma on a spatial structure. We start with the experiment performed on the lattice whose size can be compared to those in simulations and then we turn to theoretical work, experiments without spatial structure and reanalysis of other spatial experiments. Surprisingly, or not, we find that in most experiments the cooperation level declines to an asymptotic state with low but nonzero cooperation (around 20%). The only exception is pairwise dilemmas where the behaviour of subjects is qualitatively different from the cases with more players, leading to a very high cooperation level. However, the strategy updates are distinct from the most popular models of evolutionary game theory and they do not lead to the promotion of cooperation on lattices. In all experiments we notice players that use what we call "moody" conditional cooperators.

seminar schedule