Relation between human and synthetic data. The heat maps represent the per trial average reward for combinations of $αC$ (⁠$y$-axis) and $αD$ (⁠$x$-axis) in the experimental environment studied by Palminteri et al. (2017). Simulations have been performed with different softmax temperatures corresponding to the fitted temperature of the participants from that study and are averaged across 1000 agents. The stars represent the combination of fitted learning rates for each participant.