Temporal dynamics of prediction error processing during reward-based decision making

►Combination of single-trial EEG analysis and computational modeling reveals quantitative representations of prediction error. ►Cortical prediction error processing takes place in a distributed network and evolves gradually over time. ►Early representation of prediction error valence around 220 ms after feedback. ►Simultaneous late representations of prediction error valence and magnitude around 300 ms after feedback. ►Trial-by-trial neural fluctuations in prediction error representations predict future choices.