Figure 3. Mean evolutionary trajectories for the simulations of discerning recognition.

Runs that converge on the three dynamically stable equilibria C, E, or S are represented separately where n is the number of runs (out of 10000) that converge on each equilibrium and r₀ is the initial value of recognition ability r. The left column (a–c) gives mean trajectories for simulation results when r₀ is drawn from an uniform distribution over [0.5, 1.0]; the right column (d–f) gives results for r₀ = 0.5. Simulations run with p = 0.4, for equal values for help conferred or harm inflicted (b = h = 1), equal costs to confer help or inflict harm (c_h = c_b = 0.2), with mutation (μ = 0.0001).