We use a limited information environment to mimic the state of confusion in an experimental, repeated public goods game. The results show that reinforcement learning leads to dynamics similar to those observed in standard public goods games. However, closer inspection shows that individual decay...