Figure 5
Data for the Fermi rule.
Probability of changing action in the next round as a function of the payoff difference between the focal player and the best of the players that chose the opposite action. Note that this is not exactly the Fermi update rule, because it is not possible to find out which neighbor (if any) the player chose for updating her strategy; hence, choosing the best performing one is a sensible proxy for this unknown variable. The results are presented separately for the players who changed from cooperation to defection and those who changed from defection to cooperation (see legends). Results are shown for Plön (left), Madrid (center) and Zaragoza experiments (right). The solid line in the left plot (Plön) is a fit using the analytical expression of the Fermi rule (see25 for details). All error bars are the standard deviations of a binomial distribution, 
, where n is the number of samples and p is the probability of changing the action.
