Abstracts

Rationale: Non-invasive EEG recordings suggested that de-synchronizations of neural cell assemblies measured as decrease in EEG power have been associated with successful memory encoding mainly in mesial temporal structures. The existence of a frontal lobe memory network was assumed, its localization, extent and functional relevance are yet unclear. By analyzing oscillations in intracranial EEG during encoding, we aimed for the localization of verbal and non-verbal learning in the frontal lobe and evaluated the effect of resections in these areas on postoperative neuropsychological outcome.Methods: 6 patients with medically refractory epilepsy who underwent invasive EEG were tested with a memory paradigm which comprised encoding, distraction and later recognition of 100 regular nouns and faces. Encoding trials defined as one second before and two seconds after stimulus presentation were classified as hit (later remembered) and miss trials (later forgotten). For each trial and electrode we performed a time frequency analysis of the EEG. Power values were z-transformed with mean and standard deviation of all whole trials. A permutation procedure on the basis of a Student’s t-test was used to evaluate significant differences between conditions.Results: Electrode placement was left hemispheric in four and right in two patients and consisted of subdural in five and implanted depth electrodes in one case. Frontal lobe coverage was present in all, additional temporal placement in two and parietal in three patients. Memory oscillations presented as significant de-synchronizations in the alpha, beta, theta and gamma band and were mostly found 0.5 – 1.5 seconds after stimulus presentation. Isolated memory oscillations involving ≤ two adjacent electrodes were seen in all patients all over the frontal but also in the parietal and temporal lobe. Clusters of ≥ 3 neighboring electrodes involved in encoding were found in the inferior frontal gyrus (IFG) only in all four patients with IFG coverage (two left, two right hemispheric). In three patients IFG tissue revealing de-synchronization during successful encoding was resected. All of these exhibited memory deficits in 6 months postoperative testing (left resection (n=1): short term memory deficit; right resections (n=2): verbal, figural and short term memory deficit). Resections in the remaining three patients also included memory electrodes apart from the IFG without postoperative neuropsychological impairment.Conclusions: De-synchronization of EEG in a wide frequency window in the IFG appears to be necessary for successful encoding and further elucidates the electrophysiology of learning in humans. This effect did not show a clear dichotomy between verbal memory in the left and figural memory in the right hemisphere.