| Bayesian learning model | Predictions for the influence of the number of training pairs on the learning rates were made using a fixed ratio between ayo and a; for the model simulations . |
| Learning rates | To compare these results to the Mixture-of-Kalman-Filters Model, we reran the model simulations for each participant using the trial-sequence from the point at which each participant started to learn. |
| Model comparison | However, as mentioned above, model simulations themselves were performed using the trial sequence across all blocks for each individual participant. |