Combined MEG and EEG suggest a limbic source network of the P3 including retrosplenial cortex and hippocampus.

The event-related P3 is evoked by most task-relevant or salient stimuli, but its neural generators have remained controversial, limiting the integration of P3 with functional neuroimaging. Here we reevaluate the role of hippocampus and retro-splenial cortex as potential generators of the P3. Combined magneto- and electroencephalography signals were recorded during a visual oddball paradigm. Observers were instructed to respond to rare targets of a deviant shape and ignore rare non-targets of a deviant color. Source analysis was based on noise-normalized minimum-norm estimates in an individual, MRI-based source space. Critically, the cortical source space was extended to include the cornu ammonis of the hippocampus. Source analysis at the P3 peak showed strong sources in retrosplenial cortex and a weaker source in hippocampus. Activity in left somatosensory cortex was related to the button presses indicating target detection. Subsequent activity was observed in insula and anterior mid-cingulate cortex. Further stimulation studies were used to probe the spread of these sources and the plausibility of their combination to explain the data. In these models, EEG was dominated by retro-splenial cortex, while MEG was dominated by activity in primary somatosensory cortex. Except for somatosensory cortex, EEG showed better signal-to-noise ratio than MEG, but the combination of MEG and EEG increased the signal-to-noise ratio and specificity at the source level. Overall, these data support an interpretation of the P3 as physiological indicator of activity in the limbic system for target and distractor stimuli.