In a prospective study lasting 6 months, we recorded on video 108 seiz
ures with aura of 23 patients in an attempt to evaluate the mechanisms
involved in the encoding of memories. In 88 of those seizures, we als
o recorded an EEG. The percentage of auras remembered decreased signif
icantly with increasing severity of the seizures. The recollection of
auras was also significantly dependent on the ictal EEG changes during
the aura. Ninety-seven percent of the auras without EEG changes, 94%
of the auras with unilateral EEG changes, and 73% of the auras with bi
lateral EEG changes during the aura were remembered. The spread of the
ictal EEG pattern during the aura also showed a significant correlati
on with the severity of the ensuing seizure. Three patients with bitem
poral epilepsy made up a considerable proportion of those who never re
membered their aura before secondary generalized tonic-clonic seizures
(2 of 3) and of those who had a transient postictal amnesia of their
aura (2 of 3). The only patient who failed to remember a previously do
cumented isolated aura also suffered from bitemporal epilepsy. During
the second part of the study, we questioned whether information provid
ed during the history could be helpful in defining the type of epileps
y syndrome or localizing the EEG seizure pattern of the 80 patients wh
o had been admitted for presurgical epilepsy diagnosis. Localized (reg
ional, unilateral, and independent left and right lateral) EEG seizure
patterns occurred in 82% of the 51 patients with auras in their histo
ry as compared with 24% of the 17 patients who did not have auras in t
heir history (p < 0.01).