Figure 1.

Illustration of the learning model. Panel (A) illustrates the benefit and cost as functions of the investment actions. The two curves for the cost correspond to qualities q = 0 and q = 1. See Eqs (2, 3) for the formulas. Panel (B) shows simulated learning dynamics of the estimated values w_i and mean actions θ_i for an interaction between two individuals with qualities q₁ = 0 and q₂ = 1. The dynamics of w_i and θ_i are given in Eqs (7, 10). The starting point of learning was (arbitrarily) chosen as w_i = 1.0 and θ_i = 0.2. The dashed lines are one-shot game predictions for the estimated value w_i and the mean investment θ_i, corresponding to the investments in Eq. (14). These values of θ_i are also indicated in panel (A). Parameter values are: g = 2, B₀ = 1, B₁ = 4, B₂ = −2, K₁ = 1, K₁₁ = 1, K₁₂ = −1, σ = 0.05, α_w = 0.04, and α_θ = 0.002.