Fig. 3.

Mean accuracy (a) and RT (b) across participants as it develops throughout learning. Solid lines represent the mean across experimental blocks and participants, while the shaded areas represent 95% confidence intervals. Mean accuracy (c) and RT (d) across participants and for different pairs of options. Choices between options AC and BD were easier than between AB and CD, while the mean reward was highest in pair CD followed by BD, AC, and AB. The dots represent the mean across trials, while the bars represent 95% confidence intervals. Multilevel bootstrap was performed to account for repeated measures and therefore individual variability