Fraction of cooperators in every round of the three parts of the experiment (in the first and the last ones players remain in the same node of the lattice along the whole experiment, whereas in the control part players are shuffled every round). Players are arranged in a lattice with periodic boundary conditions, and play a PD game with each of its 8 surrounding neighbors in the lattice. With the notation C for cooperation, D for defection, and for the payoff obtained by a player who plays X against an opponent who plays Y, the payoff matrix of each of these PD games is cents of a euro, cents, cents, and cents. These payoffs conform a weak PD game —the most favorable to promote cooperation— because . This setup is entirely similar to that of Nowak and May's simulations [5] except for the size of the lattice (simulations are performed on lattices, with ) and the lack of self-interactions (see [17] for further comments).