The flexible updating of internal models of the environment relies on the ability to distinguish informative from uninformative surprising outcomes in changing environments. Neural responses to surprise, particularly the P300 ERP component, have been proposed to reflect attention allocation as well as the updating of internal models of the environment (Donchin, 1981; Polich, 2007). In the current study we use EEG to investigate the role of surprise signals for the updating of internal representations in two conditions: (1) a reversal learning condition in which surprise signals are informative for model updating and (2) a (perceptually identical) oddball condition in which surprise signals need to be ignored.

Behavioral data from 46 younger adults (ages 20-28) show behavioral adaption effects (increased accuracy and decreased reaction times) following informative surprising outcomes in the reversal learning condition in comparison to uninformative outcomes in the oddball condition.

An analysis of the P300 component during outcome processing revealed a larger dissociation between surprising and unsurprising (informative) outcomes in the reversal learning condition as compared to the oddball condition. Furthermore, surprise trials in the reversal condition were associated with frontal positivity, potentially reflecting prefrontal cortex engagement during internal model updating.

Overall, our findings suggest that behavioral adaptation and internal model updating following surprising informative feedback can be observed both at the behavioral, and neural level by examining ERP components such as the P300 as well as frontal activity. Future analyses aim to relate these neural dynamics to eye movements and pupil dilation.