Figure 7. U-shaped mutation kernels lead to more cooperation in high cost scenarios.

As in Fig. 6c, both graphs show the evolving cooperation rate for the space of deterministic memory-one strategies (blue) and stochastic memory-one strategies (yellow). However, here we have varied the error rate of players (ε = 1% for frequent errors, ε = 0.01% for rare errors). In addition, the cooperation probabilities p_i of new mutant strategies are now taken from a beta-distribution. The beta-distribution has the density function f(p) = Cp^α−1(1 − p)^β−1, with C being a normalization factor. The values α = β = 1 yield the uniform distribution on [0,1], as used in Fig. 6; here, we have taken α = β = 0.1, yielding a strongly U-shaped distribution. All other parameters are the same as in Fig. 6.