Fig. 6.

Evolutionary dynamics of unbending strategies in finite populations. Shown are a) the stationary abundance of IPD strategies (as indicated on the x-axis) under the limit of rare mutations and for different selection strengths β and b) the evolutionary pathways in pairwise competition dynamics (the direction of the arrows indicates dominance where the IPD strategy at the end is favored over that at the start; double arrows indicate neutral evolution). We consider a set of prescribed IPD strategies, including members belonging to unbending strategies from class A (the PSO Gambler) and from class D (generous ZD with $\chi \mathrm{\prime }=2$). TFT is an extreme boundary case of unbending strategies from class D (cf. point A in Fig. 3c). We use the Moran process for evolutionary updating and study the long-term mutation-selection equilibrium. Parameters: population size $N=100$, mutation rate $\mu \to 0$, selection strength $\beta =0.01$, 0.1, 1, payoff values: $R=3$, $S=0$, $T=5$, $P=1$.