Models comparison in experiment 1.

<p>(A) We compared the ‘Reward T’ model to all the other models by examining the paired differences in WAIC scores across models and participants. The graph presents the differences of each model’s WAIC from the ‘Reward T’ model. The ‘Reward T’ model performed significantly better than all other models in explaining participants’ choices. (B) We compared the ‘SSAT T’ model to all other models by examining the paired differences in R² scores across models and participants. The ‘SSAT T’ model gave a significantly better prediction of confidence reports than all other models except the ‘SSAT’ model. Error bars represent SEM. * p<0.05, ** p<0.005, *** p<0.0005, **** p<0.00005.