Fig. 4: Optimal learning rates and posterior predictive checks.

Posterior learning rates in relation to the optimal learning rates in the private (A) and observed (C) conditions. Orange dots represent the optimal combination between learning rates for positive and negative PE identified via simulation; red crosses indicate the posterior means of learning rates. The posterior learning rates were employed to perform posterior predictive checks for the main behavioral findings for the private (B) and observed (D) conditions. Simulated data from posteriors were analyzed in a similar fashion as the real data and the model prediction largely matched our main behavioral effect (cf. Fig. 2A).