Limbic ERPs predict

Surgery has proven quite effective in patients with pharmacoresistant temporal lobe epilepsies (TLE). Unilateral hippocampal sclerosis, in particular, is associated with positive seizure control following surgery. However, surgical relief from epileptic seizure activity may come at the expense of neuropsychological deficits. Thus, during the evaluation of patients being considered for epilepsy surgery it is necessary to estimate the potential risk of postoperative neuropsychological deficits. Since the initial reports on H.M. it has been known that bilateral hippocampectomy leads to profound amnesia.1 Numerous other studies have confirmed the importance of medial temporal structures for declarative memory processes.2–4 The known neuropsychological risks of unilateral hippocampectomy, however, seem to be confined primarily to the languagedominant hemisphere. Specifically, unilateral removal of medial temporal lobe structures in the left hemisphere often yields verbal memory deficits.5 Since postoperative performance varies from patient to patient and there are reported cases of verbal memory improvement after left temporal lobe surgery, it may be that medial temporal structures in the nondominant hemisphere contribute to performance after surgery. Several factors have been found to correlate with changes of verbal memory after left hippocampectomy, including hippocampal volumes and amount of hippocampal sclerosis.6–8 Although these studies document the neuropsychological sequelae of the removal of left medial temporal lobe structures in these patients as a group, no single factor that predicts the specific postoperative verbal memory performance in any given individual patient has yet been identified. Prognosis from Wada testing may be promising9 but has been found to be variable.10 Recordings of intracranial event-related potentials (ERPs) in TLE-patients have identified some negative potentials in the human medial temporal lobe related to registration of infrequently occurring events (e.g. MTL-P300), and others related to word processing in both anterior medial temporal lobes (e.g. AMTL-N400).11–15 We have found that AMTLN400s to words recorded in the dominant hemisphere are correlated with preoperative verbal recall performance.16 Since these potentials can be elicited in both temporal lobes, they may prove useful in predicting the functional plasticity of bilateral medial temporal lobe structures involved in various memory processes. To examine the extent to which the non-dominant temporal lobe contributes to postoperative verbal memory performance we recorded intracranial eventrelated potentials from depth electrodes within the Learning