The main finding is that learning from rewards and punishments is biased (easier acquisition of own actions from rewards, impaired unlearning of inactions after punishments) and that biased prediction errors better describe BOLD signal in cortex (pgACC, dACC, PCC) and striatum.

Importantly, EEG correlates of biased prediction errors in cortical regions (dACC, PCC) precede those of subcortical regions (striatum).