Characterizing EEG correlates of exploratory behavior
This study aims at characterizing the EEG correlates of exploratory behavior. Decision making in an uncertain environment gives rise to a conflict between two opposing needs: gathering information about the environment and exploiting this knowledge in order to optimize future decisions. Exploratory behavior has been studied using fMRI (Daw et al., Nature 06) but, to the best of our knowledge, never with EEG. This fMRI study showed bilateral activation in the frontal and parietal cortex. A model of how subjects make their decision has been built and the label of each trial derived from it. Because of the complexity of the task, the EEG correlates of the exploratory behavior are not necessarily time locked with the action. So the EEG processing methods should be designed in order to handle signals that can shift in time across trials. Assuming that the non-overlapping regions of the probability distributions of the EEG signal over time in the two conditions (exploration, exploitation) correspond to time samples where the characteristic phenomena for each condition takes place, we are able to identify peaks of EEG activity that is not necessarily time-locked with any time clue but that are relevant to each condition. Using the same experiment protocol as the fMRI study, adapted to EEG recordings, results with 9 subjects show that the bilateral frontal and parietal areas are also the most discriminant in the EEG activity. Classification of single trials has also been done and shown to be feasible when combining several frequency bands, often from 20 to 40 Hz. Identification of the frequency bands that most contribute to discriminate exploration from exploitation behavior is a definitive advantage of EEG over fMRI. Altogether, this study strongly suggests that EEG carries significant information about exploratory behavior that can be recognized in single trials.