Evolution has not yet explained why we cooperate with strangers on a daily basis. Theoreticians have suggested
that the existence of a social structure in society allows cooperative types to cluster together and help each other. We
have performed an experiment to check whether human subjects, in a laboratory controlled structure, do cooperate
more when they play a Prisoner’s Dilemma. Our results show that on a structured population people cooperate, but
only to the limited extent found in the absence of a structure. Interestingly, the reason why theories fail to predict
this is that they do not take into account that people behave idiosyncratically. Finally, we have observed that people use the same strategy all along the experiment, meaning that learning as the game proceeds occurs in a very limited extent.
