Electroencephalography and clinical Neurophysiology 108 (1998) 567–576

Event-related potentials in patients with temporal lobe epilepsy reveal

topography specific lateralization in relation to
the side of the epileptic focus

W. Lalouschek, W. Gerschlager, J. Lehrner, C. Baumgartner,

G. Lindinger, L. Deecke, W. Lang*
Department of Clinical Neurology, University of Vienna,
Währinger Gürtel 18–20, A-1097 Vienna, Austria

Accepted for publication: 21 April 1998

Abstract

Event-related potentials (ERPs) were recorded during a continuous recognition memory paradigm in patients with left-sided (LTLE;
n = 8) or right-sided temporal lobe epilepsy (RTLE; n = 6), and in healthy control subjects (n = 24). Control subjects and both patient
groups exhibited consistent OLD/NEW ERP-differences from 200–600 ms after stimulus onset. ERPs did not differ significantly between
LTLE and RTLE patients, with respect to OLD/NEW distinction or the type of presented material (verbal vs. non-verbal). However, ERP
topography showed significant differences between LTLE and RTLE patients: in lateral fronto-temporal recordings, patients showed larger
negativities contralateral to the seizure focus, whereas we found larger negativities ipsilateral to the seizure focus in parietal recordings.
Differences between the groups were significant from 300 to 600 ms post-stimulus. As a consequence, the amplitude gradient from fronto-
temporal to parietal recordings was higher on the right side in LTLE patients and on the left side in RTLE patients. Again, differences
between LTLE and RTLE patients were highly significant. We assume that ERPs reflect disturbances of a cortico-cortical network
dependent on the side of the seizure focus in temporal lobe epilepsy. Furthermore, scalp-recorded ERPs might be a useful tool in the
prediction of the side of the seizure focus in patients with temporal lobe epilepsy.  1998 Elsevier Science Ireland Ltd. All rights reserved

Keywords: Event related potential; Temporal lobe epilepsy; Recognition memory; Lateralization

1. Introduction (e.g. memory) deficits in patients with temporal lobe epi-

Event-related potentials (ERPs) have repeatedly been
investigated in patients with temporal lobe epilepsy before
or after epilepsy surgery, and in patients with other temporal
lobe lesions (Meador et al., 1987; Loring et al., 1988; Puce
et al., 1989, 1991; Puce and Bladin, 1991; Rugg et al., 1991;
Scheffers et al., 1991; Polich and Squire, 1993; Grunwald et
al., 1995; Verleger et al., 1997). These studies were moti-
vated by several assumptions: (1) ERPs might be a useful
tool in the identification of the side of the primary seizure
focus in candidates for epilepsy surgery; (2) ERPs could
help to clarify the nature of certain neuropsychological
* Corresponding author. Tel.: +43 1 40400 3117; fax: +43 1 40400 3141.
lepsy; (3) the neural substrate underlying particular ERP
components, especially the P300, is still under debate. It
might thus be helpful to study patients with precisely
defined brain lesions to further clarify this question.
Studies with intracranial recordings in patients with tem-
poral lobe epilepsy (TLE) could consistently demonstrate
diminution or absence of limbic evoked potentials ipsilat-
eral to the epileptogenic focus (Meador et al., 1987; Puce et
al., 1989). Puce and Bladin (1991) compared auditory odd-
ball scalp and limbic P300s as recorded from patients with
unilateral TLE. Limbic P300s were unilaterally absent ipsi-
lateral to the epileptogenic focus. Scalp P3s as recorded
from C3 and C4 with reference to linked earlobe-electrodes,
on the other hand, were elicited bilaterally and there was no
significant difference in amplitude between the epilepto-

0168-5597/98/$ - see front matter  1998 Elsevier Science Ireland Ltd. All rights reserved
PII S0168-5597 (98 )0 0037-9 EEP 97742
568 W. Lalouschek et al. / Electroencephalography and clinical Neurophysiology 108 (1998) 567–576
genic and non-epileptogenic sides. The authors argued that
the absence of any asymmetry of scalp recorded P300 con-
curs with postoperative data, where scalp recorded P300s
have been found to be bilaterally symmetrical in amplitude
following temporal lobectomy (Johnson and Fedio, 1986;
Johnson, 1988; Scheffers et al., 1991). Neither Puce and
Bladin (1991) nor Johnson and collaborators have used
recordings inferior and lateral to F3, C3, P3 or F4, C4, P4,
respectively. Puce and Bladin (1991) noted that the lack of a
P300 asymmetry in scalp recordings may be due to the
electrode montage selected. Rugg et al. (1990) studied
patients with TLE and recorded presurgical P300 in sphe-
noidal electrodes with a non-cephalic reference. Sphenoidal
P300s were found to be small but without asymmetry.
Smith et al. (1986) employed a recognition memory task
and recorded cognitive potentials by intracranial electrodes
placed in the medial temporal lobe in patients suffering
from complex partial seizures. They did not address the
question of hemispheric asymmetry in relation to the side
carrying the epileptogenic focus. Grunwald et al. (1995)
studied patients having unilateral TLE with intracranial
electrodes located in the medial temporal lobe (MTL), and
recorded cognitive potentials in a visual oddball paradigm
and tasks of recognition memory. P300s as well as N400s
with repeated items were found to be lateralized, with smal-
ler amplitudes in recordings ipsilateral to the epileptogenic
focus. Rugg et al. (1991) studied scalp-recorded ERPs dur-
ing a recognition memory task for visually presented words
in patients who had undergone left-sided or right-sided
lobectomy because of unilateral TLE, in patients with left-
sided or right-sided TLE, and in patients with primary gen-
eralized epilepsy (control group). EEG was recorded from
Fz, Cz, Pz, T3 and T4. Analysis of ERPs was focused on
old/new ERP effects which were asymmetric in electrode
pair T3/T4 in patients with TLE: in 8/15 control subjects
old/new ERP effects were larger from the left than the right
hemisphere; the same asymmetry occurred in 4/14 left epi-
lepsy patients, and in 9/11 right epilepsy patients. Thus,
there was some evidence that scalp-recorded cognitive
potentials may be of clinical relevance.
The purpose of the present study was to investigate the
usefulness of scalp-recorded ERPs in the detection of asym-
metries of cognitive potentials in patients with unilateral
TLE. A continuous memory recognition task was used
which included verbal items and abstract geometric figures.
Electrode montage was based on the following considera-
tions: epileptic spikes originating in the basal part of the
temporal cortex can be detected by a bipolar field pattern
with an area of negativity in inferior and lateral recordings
(FT9, T7, TP9 or FT10, T8, TP10), and an area of positivity
around the vertex. If we assume dysfunction of the basal
temporal cortex in a cognitive task, ERP effects may most
clearly be visible in those lateral and inferior recordings
when referenced to Cz. We did not use a non-cephalic elec-
trode pair as reference (e.g. a balanced sterno-vertebral
reference, Rugg et al., 1991), because we assumed that the
gradient of the electric field may be small between those
lateral recordings and the trunk. For example, Puce et al.
(1994) found only poorly visible P300s in 7/13 patients
when recording EEG from sphenoidal electrodes with refer-
ence to a balanced sterno-vertebral electrode pair (as also
used in the study by Rugg et al., 1991).
2. Subjects and methods
2.1. Subjects
2.1.1. Healthy controls
Twenty-four right-handed subjects (13 males, 11
females) ranging from 20 to 34 years of age participated
in the study. Hand dominance was assessed with a modified
version of the Edinburgh Inventory (Salmaso and Longoni,
1985). All subjects scored 100% dextrality and were paid
for completing the experiment.
2.1.2. Patients
Twenty patients suffering from medically intractable
temporal lobe epilepsy gave their informed consent to par-
ticipate in this study. All patients were investigated with
non-invasive intensive video-EEG-monitoring, and high-
resolution MRI. Most of the patients also underwent intra-
carotid sodium amobarbital test for possible epilepsy sur-
gery (Baumgartner et al., 1997). Interictal spikes were
either classified as unilateral (.90% of spikes recorded
over one side) or bilateral. Diagnosis of temporal lobe epi-
lepsy was based on typical clinical seizure semiology,
interictal as well as ictal EEG findings and the results of
MRI scan. From these investigations we were able to select
a clearly defined group of patients: only patients with uni-
lateral interictal spikes and without an extrahippocampal
lesion were investigated, thus obtaining a total of 14
patients of whom 8 had a left-sided (LTLE) and 6 a right-
sided (RTLE) seizure focus. Ten of these patients also
underwent a WADA test for the determination of language
dominance. Clinical and other (EEG-spike localization,
MRI, WADA-test) findings of these patients are presented
in Table 1.
2.2. Experimental procedure
2.2.1. ERP paradigm
Subjects viewed series of items which belonged to two
categories, verbal and nonverbal (see below). Verbal and
nonverbal items were presented in an alternated order.
Repetitions of items occurred with a probability of 25%.
There was a lag of one item between initial presentation
and repetition. The interval between the presentation of
two items was 2.1 s. Seven series of 100 items and one
consisting of 68 items were presented (768 items in total).
Subjects were instructed to make decisions upon each item
by lifting the two index fingers either once (‘new’) or twice
 W. Lalouschek et al. / Electroencephalography and clinical Neurophysiology 108 (1998) 567–576 569

Table 1 trode potentials (Bauer et al., 1989). Scratching of the skin

Patients’ characteristics of the present study at recording sites reduced electrode impedance at 5 Hz to
less than 2 kQ. A computer-based, multi-channel DC ampli-
Patient no. MRI findings Language lateraliza-
tion (WADA test)
fier with digital filtering was used (Lindinger et al., 1991).
Horizontal EOG (lateral orbital rim of the right eye vs.
Left temporal lobe epilepsy lateral orbital rim of the left eye) and vertical EOG (upper
1 Left hc atrophy Left
2 Left hc atrophy Left
vs. lower orbital rim of the right eye) were recorded. Data
3 Left hc atrophy Left were digitized at a rate of 250 samples/s using a bandpass
4 Left hc atrophy Left ranging from DC to 100 Hz. All data were visually con-
5 Cavernoma left hc Bilateral trolled for artifacts and affected epochs were excluded. A
6 Left hc atrophy Bilateral linear regression was applied during off-line analysis, in
7 Left hc atrophy n.p.
8 Left hc atrophy n.p.
order to remove artifacts in the EEG resulting from eye
blinks.
Right temporal lobe epilepsy ERPs were recorded during a period of 2100 ms (1300 ms
9 Normal n.p. before and 800 ms after stimulus onset). Baseline measure-
10 Right hc atrophy Left ments were calculated from the time period 300 ms before
11 Right hc atrophy Bilateral
12 Right hc atrophy Bilateral
stimulus onset.
13 Ganglioglioma right hc Left
14 Right hc atrophy n.p. 2.5. ERP analysis
n.p., not performed; hc, hippocampus.
Mean amplitudes were calculated in 100 ms epochs
between 100 and 600 ms after stimulus onset, thus obtaining
(‘old’), and thereby crossing a light barrier. They had to 5 epochs (N1–N5). ERP differences between groups were
respond fast but as accurately as possible. calculated using analysis of variance (ANOVA) in two
steps: step 1, comparison of ERPs between healthy controls
2.2.2. Stimuli and temporal lobe epilepsy patients; step 2, comparison of
Stimulus display on a computer screen subtended an ERPs between patients with a left- and a right-sided seizure
angle of 2 degrees in the horizontal and the vertical plane, focus, representing the main goal of the present study.
stimulus duration was 200 ms. Stimuli were either verbal The following ANOVA design was used. The between-
(pronounceable non-words) or nonverbal (spatial patterns). subject factor was GROUP (step 1, healthy controls vs.
Non-words were created by a string of 3 letters (consonant- temporal lobe epilepsy patients; step 2, patients with left
vowel-consonant) and were presented white on a black temporal vs. patients with right temporal lobe epilepsy;
background. Figures were abstract, geometric patterns. LTLE vs. RTLE). Within-subject factors were:
They consisted of small bright triangles and quadrangles
on a black background (Beisteiner et al., 1996).

2.3. Electrode montage

In order to compare lateralization and topography of ERPs

between patients with a left-sided or right-sided epilep-
togenic focus, we applied an electrode montage in-
cluding electrode sites known to record from the basal
temporal lobe using Cz as reference. ERPs were recor-
ded from 18 sites: F7, F8, FT9, FT10, FC5, FC6, T7, T8,
TP9, TP10, CP5, CP6, P7, P8, P3, P4, O1, O2 (Fig. 1). The
terminology of Sharbrough et al. (1991) was used by
which electrode T3 is labeled T7, T4 is labeled T8, T5
becomes P7 and T6 becomes P8. By doing so, all electrodes
at a given horizontal position have the same numeric sub-
script.

2.4. ERP recording

Non-polarizable Ag/AgCl electrodes were connected to

the scalp via salt bridges (silicon rubber filled with electrode
gel), in order to reduce skin potential and to stabilize elec- Fig. 1. Electrode montage in the present study.
570 W. Lalouschek et al. / Electroencephalography and clinical Neurophysiology 108 (1998) 567–576

OLD/NEW (2 levels) Table 3

MATERIAL (VERBAL/NONVERBAL; 2 levels)
Mean reaction times (in s) of control subjects and patients for verbal and
TOPO (9 levels) nonverbal stimuli (standard error in parentheses)
RIGHT/LEFT (2 levels)
Control subjects Patients
Only trials with correctly identified old/new stimuli were
Verbal material 0.70 (0.02) 0.73 (0.04)
used for the analysis. Effects of within-subject factors OLD/ Nonverbal material 0.75 (0.02) 0.81 (0.06)
NEW, RIGHT/LEFT, and TOPO were tested in an ANOVA
design for amplitudes of each time period (N1–N5). Ana-
performance for old compared to new stimuli (F = 46.69;
lyses of topographical effects (TOPO) were based on nor-
P , 0.001) and for nonverbal compared to verbal stimuli
malized data according to McCarthy and Wood (1985).
(F = 98.51; P , 0.001). The interactions old/new by group
Greenhouse-Geisser epsilon was used for the correction of
(control subjects vs. patients) and material (verbal vs. non-
P values in repeated analyses. When the global ANOVA
verbal) by group were not significant. Comparison of the
indicated a significant difference between the two groups
patient groups (LTLE vs. RTLE patients) revealed a some-
for one or more of the within-subject factors, the respective
what poorer memory performance in LTLE patients, how-
effects were tested using univariate analyses. A significant
ever the difference did not reach significance. Similarly, the
interaction TOPO × RIGHT/LEFT indicated a right/left
interactions old/new by group (LTLE vs. RTLE patients)
ERP difference at particular recording sites and justified a
and material by group were not significant.
subsequent analysis of RIGHT/LEFT-ERP differences in
Reaction times of control subjects and patients did not
single pairs of electrodes. This was accomplished by com-
differ significantly. Control subjects and patients showed
paring difference amplitudes of single electrode pairs (e.g.
longer reaction times for nonverbal than for verbal stimuli
F8 minus F7) with Student’s t test (two-tailed) between the
(F = 28.85; P , 0.001). Differences between groups (con-
two groups (step 1, control subjects vs. patients; step 2,
trol subjects vs. patients) were not significant (Table 3).
LTLE vs. RTLE patients). On the other hand, a significant
There were no global or material-specific differences in
interaction GROUP × RIGHT/LEFT in the absence of a
reaction times between LTLE and RTLE patients.
significant interaction between TOPO and RIGHT/LEFT
indicated a global RIGHT/LEFT difference of ERPs
3.2. ERP data
between the groups.
3.2.1. ERPs in control subjects and patients
2.6. Behavioral analysis
Control subjects exhibited a consistent OLD/NEW effect
Reaction times (RT) and frequencies of correct and incor-
rect decisions upon old/new distinction were calculated and
compared between groups by means of analysis of variance
(ANOVA) (step 1, healthy controls vs. patients; step 2,
LTLE vs. RTLE patients).

3. Results

3.1. Behavioral data

Comparison of control subjects and patients revealed a

significant difference between groups, in that patients gave
more incorrect answers (F = 6.49; P , 0.02) (Table 2).
Both control subjects and patients showed a poorer memory

Table 2

Mean percentage of incorrect answers/stimulus category in control sub-

jects and patients (standard error in parentheses)

Control subjects LTLE RTLE

All 13.6 (1.3) 21.1 (2.5) 16.5 (1.4)

Verbal/new 3.0 (0.8) 4.5 (1.3) 4.1 (1.7))
Verbal/old 17.7 (2.7) 30.3 (5.1) 20.1 (3.1)
Nonverbal/new 6.3 (1.2) 8.6 (2.2) 7.0 (1.8) Fig. 2. Averages for control subjects, LTLE and RTLE patients for cor-
Nonverbal/old 27.5 (3.1) 40.9 (5.3) 33.9 (5.7) rectly identified new (solid line) versus correctly identified old (dashed
line) stimuli (verbal and nonverbal stimuli together). Negativity up.
 W. Lalouschek et al. / Electroencephalography and clinical Neurophysiology 108 (1998) 567–576 571

Table 4
ANOVA results (P values) for the comparison between control subjects and the patient group as a whole for the factors OLD/NEW, TOPO, MATERIAL and
RIGHT/LEFT (multiple interactions not shown)

100–200 ms 200–300 ms 300–400 ms 400–500 ms 500–600 ms

GROUP P = 0.951 0.108 0.322 0.776 0.883

F = 0.00 2.69 1.00 0.08 0.02
OLD/NEW P = 0.733 0.001 0.000 0.000 0.001
F = 0.12 12.86 28.13 19.14 13.05
MATERIAL P = 0.000 0.003 0.720 0.989 0.006
F = 29.63 9.62 0.13 0.00 8.46
TOPO P = 0.000 0.000 0.000 0.000 0.000
F = 61.65 44.09 47.56 49.01 50.57
RIGHT/LEFT P = 0.021 0.887 0.716 0.170 0.292
F = 5.71 0.02 0.13 1.95 1.14
GROUP × OLD/NEW P = 0.625 0.273 0.234 0.350 0.680
F = 0.24 1.23 1.46 0.89 0.17
GROUP × MATERIAL P = 0.154 0.530 0.290 0.411 0.454
F = 2.10 0.40 1.15 0.69 0.57
GROUP × TOPO P = 0.358 0.499 0.729 0.748 0.625
F = 1.05 0.74 0.36 0.32 0.51
GROUP × RIGHT/LEFT P = 0.845 0.679 0.170 0.092 0.157
F = 0.04 0.17 1.94 2.96 2.08

Topographical analyses are based on normalized data.

from 200–600 ms post-stimulus when ERPs for correctly
identified old items were significantly more negative-going
than for correctly identified new items (Fig. 2).
The general course of ERPs in patients was similar to that
in control subjects, including a consistent OLD/NEW ERP
effect in both patient groups (Fig. 2). There were no signifi-
cant differences between control subjects and the patient
group as a whole concerning the OLD/NEW effect or mate-
rial specific effects (MATERIAL) on ERPs (Fig. 2, Table
4).
in fronto-temporal recordings and a larger negativity on the
left side in parietal recordings, whereas an opposite pattern
was found in RTLE patients.
In a subsequent step of analysis, the relation between
fronto-temporal and parietal electrodes was analyzed by
calculating the fronto/temporo-parietal gradient (S; Subtrac-
tion) (right hemisphere: gradient SR: (F8 + FT10)/2 − P4;
left hemisphere: gradient SL: (F7 + FT9)/2 − P3). Both

3.2.2. ERPs in left vs. right temporal lobe epilepsy patients

(LTLE vs. RTLE)
LTLE and RTLE patients exhibited similar OLD/NEW
ERP effects, both in verbal and nonverbal stimuli (Fig. 2).
Accordingly, a global ANOVA revealed no significant dif-
ferences between LTLE and RTLE patients concerning the
factors OLD/NEW and MATERIAL (Table 5).
A significant difference of ERP lateralization was found
between the patient groups from 300–600 ms post-stimu-
lus, which was topography specific (interaction GROUP ×
TOPO × RIGHT/LEFT) (Table 5). Further analyses re-
vealed that the LTLE group exhibited an ERP lateralization
with negative amplitudes being larger in right fronto-tem-
poral recordings as compared to left fronto-temporal record-
ings (F7/8, FT9/10) (i.e. lateralization to the right), whereas
RTLE patients showed a lateralization to the left (Fig. 3,
Table 6). At the same time, a completely opposite effect
could be observed over P3/4, where LTLE patients showed
a lateralization to the left and RTLE patients to the right
Fig. 3. Averages for control subjects, LTLE, and RTLE patients across all
(Fig. 3, Table 6). Similar effects were observed over P7/8 conditions (old, new, verbal, nonverbal; only trials with correct stimulus
and CP5/6, but did not reach significance. In summary, identification). Left-sided (solid line) versus right-sided (dashed line) elec-
LTLE patients showed a larger negativity on the right side trodes. Negativity up.
572 W. Lalouschek et al. / Electroencephalography and clinical Neurophysiology 108 (1998) 567–576

Table 5
ANOVA results (P values) for the comparison between LTLE and RTLE patients

100–200 ms 200–300 ms 300–400 ms 400–500 ms 500–600 ms

GROUP P = 0.188 0.110 0.412 0.907 0.973

F = 1.92 2.91 0.72 0.01 0.00
GROUP × OLD/NEW P = 0.256 0.189 0.978 0.150 0.222
F = 1.40 1.90 0.00 2.32 1.63
GROUP × MATERIAL P = 0.126 0.363 0.842 0.566 0.927
F = 2.64 0.88 0.04 0.34 0.01
GROUP × RIGHT/LEFT P = 0.780 0.329 0.390 0.213 0.117
F = 0.08 1.02 0.79 1.70 2.80
GROUP × OLD/NEW × MATERIAL P = 0.293 0.709 0.152 0.559 0.631
F = 1.19 0.14 2.30 0.36 0.24
GROUP × OLD/NEW × RIGHT/LEFT P = 0.452 0.716 0.256 0.622 0.999
F = 0.60 0.14 1.40 0.25 0.00
GROUP × MATERIAL × RIGHT/LEFT P = 0.567 0.733 0.886 0.587 0.651
F = 0.34 0.12 0.02 0.31 0.21
GROUP × OLD/NEW × P = 0.690 0.447 0.836 0.233 0.535
MATERIAL × RIGHT/LEFT
F = 0.17 0.61 0.04 1.55 0.40
GROUP × TOPO P = 0.448 0.666 0.955 0.730 0.482
F = 0.81 0.45 0.05 0.27 0.74
GROUP × OLD/NEW × TOPO P = 0.928 0.639 0.814 0.943 0.803
F = 0.15 0.61 0.28 0.11 0.22
GROUP × MATERIAL × TOPO P = 0.765 0.769 0.717 0.705 0.716
F = 0.32 0.23 0.28 0.36 0.48
GROUP × TOPO × RIGHT/LEFT P = 0.351 0.306 0.037 0.015 0.024
F = 1.13 1.24 2.81 3.85 3.60
GROUP × OLD/NEW × P = 0.506 0.353 0.732 0.608 0.700
MATERIAL × TOPO
F= 0.80 1.09 0.41 0.52 0.46
GROUP × OLD/NEW × TOPO × P= 0.993 0.581 0.202 0.161 0.408
RIGHT/LEFT
F= 0.05 0.61 1.59 1.79 0.99
GROUP × MATERIAL × TOPO × P= 0.330 0.515 0.489 0.468 0.590
RIGHT/LEFT
F= 1.18 0.72 0.79 0.79 0.51
GROUP × OLD/NEW × P= 0.852 0.850 0.479 0.581 0.523
MATERIAL × TOPO × RIGHT/LEFT
F= 0.20 0.26 0.87 0.62 0.70

The interaction GROUP × RIGHT/LEFT × TOPO is significant from 300–600 ms post-stimulus. Topographical analyses are based on normalized data.

patient groups showed larger fronto-temporal than parietal
negativities on both sides. But the relation of the gradients
SL and SR (S1: SR minus SL) was completely opposite in
the two patient groups, thereby reflecting the above
described right-left differences between LTLE and RTLE
patients: whereas LTLE patients showed a greater fronto/
temporo-parietal gradient on the right side, RTLE patients
showed a significantly greater fronto/temporo-parietal gra-
dient on the left side. The difference between LTLE and
RTLE patients was continuously increasing from 100–600
ms post-stimulus and was highly significant for all time
periods after 300 ms post-stimulus (Fig. 4, Table 6).
The described differences between the two patient groups
were also very consistent on a single subject basis, allowing
a distinction between LTLE and RTLE patients on the basis
of the right to left relation in fronto-temporal and parietal
recordings as well as of the fronto/temporo-parietal gradi-
ents in 6/8 LTLE and 6/6 RTLE patients (Fig. 5). Interest-
ingly, none of the described effects and differences was
material-specific.
4. Discussion
4.1. Behavioral data
The fact that patients with TLE had an overall poorer
memory performance was not unexpected, however the lit-
erature suggests rather material-specific than global mem-
ory impairment, particularly for words in patients with
LTLE (Hermann et al., 1997). This may be due to the rela-
tively small number of patients in our study. Furthermore,
the focus of our study was on ERPs rather than on neurop-
sychological aspects. Therefore, the experimental paradigm
might not be sensitive enough to detect material specific
behavioral deficits in our patients. The observation that all
 W. Lalouschek et al. / Electroencephalography and clinical Neurophysiology 108 (1998) 567–576 573

Table 6
Mean amplitude differences (mV) between electrode pairs and for the difference between gradients SR/SL in LTLE and RTLE patients

LTLE patients RTLE patients

100– 200– 300– 400– 500– 100– 200– 300– 400– 500–
200 ms 300 ms 400 ms 500 ms 600 ms 200 ms 300 ms 400 ms 500 ms 600 ms

F8 − 7a mV −0.610 −1.184 −1.011 −2.588* −3.233** 0.224 0.425 1.858 2.757* 3.408**
P= 0.029 0.205 0.445 0.129 0.069 0.694 0.518 0.016 0.026 0.038
FC6 − 5 mV 0.372 −0.370 0.210 −0.325 −0.666* 0.696 0.509 2.134 2.876 3.114*
P= 0.526 0.574 0.817 0.767 0.486 0.360 0.718 0.117 0.048 0.059
FT10 − 9 mV 0.333 −0.379* −0.289** −0.423** −0.563* 0.647 1.333* 2.329** 2.600** 2.662*
P= 0.353 0.270 0.582 0.489 0.321 0.158 0.098 0.020 0.025 0.051
TP10 − 9 mV −0.158 −0.925 −0.902* −0.983 −0.888 0.274 0.679 1.076* 0.583 0.582
P= 0.565 0.127 0.171 0.183 0.121 0.415 0.392 0.079 0.304 0.303
T8 − 7 mV 0.284 −0.893 −0.928 −1.216 −1.197 0.898 0.811 1.969 2.337 2.581
P= 0.483 0.229 0.352 0.302 0.316 0.258 0.630 0.277 0.176 0.200
CP6 − 5 mV 2.143 1.199 2.012 1.674 1.457 0.818 0.445 0.318 −0.126 −0.331
P= 0.069 0.283 0.058 0.123 0.130 0.312 0.814 0.827 0.916 0.809
P4 − 3 mV 0.973 0.267 0.780* 0.635* 0.724** −1.625 −1.255 −1.984* −2.915* −3.080**
P= 0.195 0.706 0.257 0.403 0.350 0.288 0.552 0.160 0.048 0.035
P8 − 7 mV 1.015 0.447 1.313 0.903 0.140 1.488 2.792 1.856 −0.142 −0.957
P= 0.404 0.796 0.491 0.666 0.959 0.277 0.142 0.249 0.920 0.385
O2 − 1 mV 1.115 1.625 1.480 1.340 1.202 0.116 1.584 1.040 0.107 −0.152
P= 0.052 0.095 0.130 0.249 0.326 0.803 0.046 0.176 0.908 0.900
SR − SLb mV −1.112 −1.048 −1.430** −2.140** −2.622** 2.061 2.134 4.077** 5.593** 6.116**
P= 0.125 0.147 0.096 0.063 0.107 0.249 0.267 0.014 0.007 0.006
a
Amplitude differences: positive values, larger negativity over the left-sided electrode (i.e. lateralization to the left); negative values, larger negativity over
the right-sided electrode (i.e. lateralization to the right).
b
Gradient difference SR − SL: positive values, larger gradient over the left hemisphere; negative values, larger gradient over the right hemisphere.
P values refer to right/left-amplitude differences within the groups. *Significantly different between LTLE and RTLE patients at the 5% level; **significantly
different between LTLE and RTLE patients at the 1% level.

groups showed poorer memory scores for old as compared
to new stimuli is probably due to a conservative bias of our
subjects.
4.2. ERPs
ERPs recorded intracranially from the mesial temporal
lobe are a sensitive and specific predictor for the side of
the primary seizure focus in the pre-surgical evaluation of
patients with temporal lobe epilepsy (Meador et al., 1987;
Puce et al., 1989; Puce and Bladin, 1991). In TLE, intracra-
nial temporal recordings are used in the pre-surgical analy-
sis when: (i) the hemisphere or the site of the epileptogenic
focus remains unclear after a thorough analysis of the clin-
ical and EEG seizure type, as assessed by prolonged mon-
itoring by video and surface EEG (including sphenoidal
electrodes); and (ii) there is no brain pathology by magnetic
resonance tomography (MRT) or by functional imaging
(e.g. positron emission tomography, PET) (So, 1991;
Wyler, 1991). Most patients with unilateral TLE do not
require invasive recordings. If relevant ERP data can only
be obtained by intracranial recordings, its use would be
quite limited. For that reason, Puce and Bladin (1991) com-
pared scalp-recorded P300s with intracranially recorded
P300s in patients with unilateral TLE. Although P300s
were missing in intracranial recordings of the medial tem-
poral lobe ipsilateral to the epileptogenic focus, there was
no asymmetry in scalp-recorded P300s. But this negative
finding was of limited validity because of the fact that
EEG was only recorded in C3 and C4. Rugg et al. (1991)
described a reduced old/new ERP effect in a recognition
memory task ipsilateral to the seizure focus in patients
with TLE. This finding may concur with reduced amplitudes
of old/new ERP word repetition effects, as observed by
intracranial recordings in the medial temporal lobe near
the epileptogenic focus (Elger et al., 1997). In the present
study, amplitudes of scalp-recorded cognitive potentials
were asymmetric depending on the side of the epileptogenic
focus.
Several factors are likely to explain the discrepancy
between the negative results in the study of Puce and Bladin
(1991) and the positive results of our study. (1) Puce and
Bladin (1991) used only two lateral recordings, C3 and C4.
In our study ERP asymmetries were found not only in lateral
and inferior fronto-temporal recordings (F7/8; FT9/10) but
also in parietal recordings P3/4. Most interestingly, the
asymmetry was different between the two areas. (2) Linked
earlobe-electrodes close circuit currents of the two areas and
produce a distortion of the electric fields. This reference is
likely to reduce asymmetries of brain potentials in temporal
recordings. In order to compensate for this problem, recent
studies usually interpose a resistor between linked earlobe-
electrodes. (3) Puce and Bladin (1991) employed a standard
P300 oddball paradigm, whereas we used a memory task.
574 W. Lalouschek et al. / Electroencephalography and clinical Neurophysiology 108 (1998) 567–576
(4) It is likely that there are differences in the patient groups
since invasive recordings were necessary in pre-surgical
analysis of the patients studied by Puce and Bladin (1991)
whereas non-invasive monitoring and (functional) brain
imaging was sufficient in the pre-surgical analysis of our
patients. Furthermore, the presence or absence of a morpho-
logical brain lesion and the site of the lesion (hippocampus,
extrahippocampal, temporo-lateral) was only given for a
subset of patients. (5) Puce and Bladin (1991) measured
peak amplitudes of the P300s. As done by Rugg et al.
(1991), we measured mean amplitudes of fixed time inter-
vals which may be more objective and less affected by
residual noise.
Interestingly, ERPs from healthy control subjects and
patients with temporal lobe epilepsy (LTLE and RTLE
patients together) did differ neither on a global level nor
in relation to any of the investigated parameters. Rugg et
al. (1991), who used a different experimental paradigm,
found smaller old/new ERP effects ipsilateral to the seizure
focus in patients with TLE. In contrast, we could not find
significant ERP differences between controls and patients
related to old/new distinction, nor could we demonstrate any
consistent topographical or material-specific ERP differ-
ences between these groups. However, the size of the asym-
metries in Rugg’s study was small and not present in every
patient. Furthermore, Rugg et al. suggested that an attenua-
tion of the old/new effect would be expected in those
patients with pathology in temporal lobe structures other
than the hippocampus or, alternatively, in patients with par-
ticularly severe hippocampal lesions. The discrepancy
between Rugg’s and our findings may therefore be due, at
least partly, to the fact that the patients in our study may
have suffered less severe morphological brain alterations,
since we selected only patients with hippocampal atrophy or
normal MRI-findings. On the other hand, the absence of an
ERP difference related to old/new distinction between con-
trols and patients confirms the conclusion of Rugg et al.
(1991) that the temporal lobe is not the principal generator
of these ERP effects. Recently, Rugg et al. (1996) described
a ‘parietal old/new effect’ associated with the recollection
of associative information which they attributed to retrieval
operations supported by the medial temporal lobe system.
An experiment focusing on the recollection of associative
information in patients with temporal lobe epilepsy might
thus be an interesting undertaking.
The fact that we did not find material-specific differences
in ERPs between LTLE and RTLE patients parallels the
behavioral data and may also be explained by the patient
selection, the relatively small number of patients and the
fact that the task we used in the experiment might not be
sensitive enough to detect such material-specific differences
between patient groups. However, significant and consistent
ERP differences were found from 300 to 600 ms post-sti-
mulus, between LTLE and RTLE patients which were
related to the side of the seizure focus. These differences
were topography-specific, as indicated by the interaction
GROUP × RIGHT/LEFT × TOPO, but were not affected
by old/new effects on ERPs or the type of the presented
material. Firstly, the patients showed larger negativities on
the side contralateral to the seizure focus over fronto-tem-
poral recordings (particularly FT9/10 and F7/8): patients
with a left-sided focus lateralized to the right side, whereas
patients with a right-sided focus showed an opposite pattern.
(This effect was somewhat more pronounced in F7/8 in
LTLE patients and in FT9/10 in RTLE patients. However,
differences between groups were significant for both elec-
trode pairs.) Most of the previous studies on scalp-recorded
ERPs had failed to demonstrate a relationship between ERP
lateralization and the side of the seizure focus in patients
with temporal lobe epilepsy (Rugg et al. (1991) reported an
asymmetry in TLE patients only for the old/new effect). Our
findings may primarily be due to the fact that we recorded
from a larger number of electrodes, thus obtaining more
information about topographical cortical activity compared
to earlier studies.
Considering the possible underlying mechanism of the
observed ERP differences, the most obvious answer
would be that the temporal lobe ipsilateral to the seizure
focus simply generates somewhat lower amplitudes. Since
a morphological lesion of the hippocampal region was pre-
sent in most of our patients (mainly hippocampal atrophy),
this finding alone would not be too surprising. However, the
ERP differences between LTLE and RTLE patients were
not restricted to the fronto-temporal recordings. Surpris-
ingly, we found a reversal of this effect over parietal elec-
trodes P3/4 where the patients showed larger amplitudes
ipsilateral to the seizure focus (a similar tendency was
observed over CP5/6 and P7/8, particularly in LTLE
patients, but did not reach significance). Lateralization
over P3/4 thereby was inversely correlated to the lateraliza-
tion over fronto-temporal recordings.
There are two main aspects with respect to the described
differences in ERP topography between LTLE and RTLE
patients. On the one hand, temporal lobe epilepsy patients
exhibit decreased negativity ipsilateral to the seizure focus
over fronto-temporal recordings. On the other hand, they
show increased negativity over parietal recordings on the
same side. Both effects are increasing with time after sti-
mulus presentation. As a consequence of the above
described processes, the fronto/temporo-parietal gradient
on the left side (SL; pooled F7/FT9 minus P3) was much
larger than on the right side (gradient SR: pooled F8/FT10
minus P4) in RTLE patients which was increasing with time
post-stimulus, whereas the gradient was larger on the right
than on the left sides in LTLE patients. Accordingly, the
gradient difference between SR and SL (S1) showed a com-
pletely opposite behavior in both patient groups (Fig. 5).
The described differences between LTLE and RTLE
patients were also very consistent on an individual level
(Fig. 5). These results suggest that scalp-recorded ERPs
may be sensitive to the side of the seizure focus, provided
a suitable electrode montage and the inclusion of later ERP
 W. Lalouschek et al. / Electroencephalography and clinical Neurophysiology 108 (1998) 567–576
intervals in the analysis. Investigations in a larger number of
patients will help to define the topographical differences
between LTLE and RTLE patients more exactly and to
obtain a measure of the specificity and sensitivity of ERP
topography in the identification of the side of the seizure
focus in temporal lobe epilepsy.
The underlying mechanism of the topographical ERP dif-
ferences between LTLE and RTLE patients cannot be sim-
ply inferred from the results of our study. However, it is
known from neuroanatomical studies that the medial and
anterior temporal lobe as well as the posterior parietal
lobe are interconnected with several areas of the frontal
lobe, including the dorsolateral frontal cortex (Brodmann’s
areas 9, 45, and 46) (Pandya and Yeterian, 1985). Within
this network, the parietal association cortex is probably
under inhibitory control of the prefrontal cortex (Lhermitte,
1983; Lhermitte et al., 1986). Accordingly, decreased nega-
tivity over fronto-temporal regions may be accompanied by
increased parietal negativity ipsilateral to the seizure focus
in the present study. This interaction was also reflected in
the relation of the fronto-parietal gradients on the right and
on the left side which was significantly different between
LTLE and RTLE patients. The effect was highly consistent
in individual patients with LTLE or RTLE. Despite the
relatively small number of patients, these results suggest
differences in functional brain organization in relation to
the side of the seizure focus.
In conclusion, the present study provided evidence that
scalp-recorded ERPs can be a useful tool in the prediction of
the side of the seizure focus in patients with temporal lobe
epilepsy, provided a suitable electrode and reference choice
and the inclusion of adequate time epochs. Furthermore, the
presented results indicate differences in brain dynamics
between patients with a left- or a right-sided seizure
focus, which may be a consequence of disturbed stimulus
processing on the side of the lesion. These differences seem
not to be correlated with overt behavioral differences though
our study did not focus on the correlation of ERP effects
with neuropsychological parameters. Thus it may be that the
ERP differences are reflected in more subtle neuropsycho-
Fig. 4. Mean gradient difference S1 (difference between the fronto/tem-
poro-parietal gradients SR and SL) in control subjects, LTLE and RTLE
patients. Positive values indicate a larger fronto/temporo-parietal gradient
on the left side.
575
Fig. 5. Gradient difference S1 (difference between the fronto/temporo-
parietal gradients SR and SL) in individual LTLE and RTLE patients.
Positive values indicate a larger fronto/temporo-parietal gradient on the
left side.
logical differences between LTLE and RTLE patients. We
are now studying a larger group of patients with the
described paradigm and electrode montage, in order to elu-
cidate the underlying mechanism of ERP differences
between LTLE and RTLE patients in more detail and to
examine possible relationships with behavioral performance
and with other measures of regional brain activity
(HMPAO-SPECT). At the same time, we are performing a
prospective evaluation of the predictive power of scalp-
recorded ERPs with respect to the side of the seizure
focus in single patients with temporal lobe epilepsy.
Acknowledgements
We wish to thank Ingrid Klicpera for technical assistance.
The study was supported by the FWF (Fonds zur Förderung
der Wissenschaftlichen Forschung; P8215-MED).
References
Bauer, H., Korunka, C. and Leodolter, M. Technical requirements for
high-quality scalp DC recordings. Electroenceph. clin. Neurophysiol.,
1989, 72: 545–547.
Baumgartner, C., Czech, T., Feucht, M., Schindler, E. and Podreka, I.
Presurgical diagnosis of epilepsy and surgical epilepsy treatment.
Wien. Klin. Wochenschr., 1997, 109: 180–191.
Beisteiner, R., Huter, D., Edward, V., Koch, G., Franzen, P., Egkher, A.,
Lindinger, G., Baumgartner, C. and Lang, W. Brain potentials with old/
new distinction of non-words and geometric figures. Electroenceph. clin.
Neurophysiol., 1996, 99: 517–526.
Elger, C.E., Grunwald, T., Lehnertz, K., Kutas, M., Helmstaedter, C.,
Brockhaus, A., Vanroost, D. and Heinze, H.J. Human temporal lobe
potentials in verbal learning and memory processes. Neuropsychologia,
1997, 35: 657–667.
Grunwald, T., Elger, C.E., Lehnertz, K., Van Roost, D. and Heinze, H.J.
Alterations of intrahippocampal cognitive potentials in temporal lobe
epilepsy. Electroenceph. clin. Neurophysiol., 1995, 95: 53–62.
Hermann, B.P., Seidenberg, M., Schoenfeld, J. and Davies, K. Neuropsy-
576 W. Lalouschek et al. / Electroencephalography and clinical Neurophysiology 108 (1998) 567–576
chological characteristics of the syndrome of mesial temporal lobe
epilepsy. Arch. Neurol., 1997, 54: 369–376.
Johnson, R.J. Scalp recorded P300 activity in patients following unilateral
temporal lobectomy. Brain, 1988, 111: 1517–1529.
Johnson, R.J. and Fedio, P. P300 activity in patients following unilateral
temporal lobectomy: a preliminary report. Electroenceph. clin. Neuro-
physiol., 1986, 38 (suppl): 552–554.
Lhermitte, F. ‘Utilization behaviour’ and its relation to lesions of the
frontal lobes. Brain, 1983, 106: 237–255.
Lhermitte, F., Pillon, B. and Serdaru, M. Human autonomy and the frontal
lobes. Part I: Imitation and utilization behavior: a neuropsychological
study of 75 patients. Ann. Neurol., 1986, 19: 326–334.
Lindinger, G., Svassek, P.L.W. and Deecke, L. PC-supported 64-channel
DC-EEG amplifier. In: K.P. Adlassnig, G. Grabner, S. Bengtsson and R.
Hansen (Eds.), Medical Informatics Europe 1991 Proceedings, Vienna,
Austria, August 19–22. Springer, Berlin, Heidelberg, 1991, pp. 1005–
1009.
Loring, D.W., Meador, K.J., King, D.W., Gallagher, B.B., Smith, J.R. and
Flanigin, H.F. Relationship of limbic evoked potentials to recent mem-
ory performance. Neurology, 1988, 38: 45–48.
McCarthy, G. and Wood, C.C. Scalp distributions of event-related poten-
tials: an ambiguity associated with analysis of variance models. Elec-
troenceph. clin. Neurophysiol., 1985, 62: 203–208.
Meador, K.J., Loring, D.W., King, D.W., Gallagher, B.B., Gould, M.J.,
Flanigin, H.F. and Smith, J.R. Limbic evoked potentials predict site of
epileptic focus. Neurology, 1987, 37: 494–497.
Pandya, D.N. and Yeterian, E.H. Architecture and connections of cortical
association areas. In: A. Peters and E.G. Jones (Eds.), Cerebral cortex.
Plenum, New York, 1985, pp. 3–61.
Polich, J. and Squire, L.R. P300 from amnesic patients with bilateral
hippocampal lesions. Electroenceph. clin. Neurophysiol., 1993, 86:
408–417.
Puce, A., Kalnins, R.M., Berkovic, S.F., Donnan, G.A. and Bladin, P.F.
Limbic P3 potentials, seizure localization, and surgical pathology in
temporal lobe epilepsy. Ann. Neurol., 1989, 26: 377–385.
Puce, A., Andrewes, D.G., Berkovic, S.F. and Bladin, P.F. Visual recogni-
tion memory. Neurophysiological evidence for the role of temporal
white matter in man. Brain, 1991, 114: 1647–1666.
Puce, A. and Bladin, P.F. Scalp and limbic P3 event-related potentials in
the assessment of patients with temporal lobe epilepsy. Epilepsia, 1991,
32: 629–634.
Puce, A., Berkovic, S.F., Cadusch, P.J. and Bladin, P.F. P3 latency jitter
assessed using 2 techniques. II. Surface and sphenoidal recordings in
subjects with focal epilepsy. Electroenceph. clin. Neurophysiol., 1994,
92: 555–567.
Rugg, M.D., Roberts, R.C., Potter, D.D., Nagy, M.E. and Pickles, C.D.
Endogenous event-related potentials from sphenoidal electrodes. Elec-
troenceph. clin. Neurophysiol., 1990, 76: 331–338.
Rugg, M.D., Roberts, R.C., Potter, D.D., Pickles, C.D. and Nagy, M.E.
Event-related potentials related to recognition memory. Effects of uni-
lateral temporal lobectomy and temporal lobe epilepsy. Brain, 1991,
114: 2313–2332.
Rugg, M.D., Schloerscheidt, A.M., Doyle, M.C., Cox, C.J.C. and Patching,
G.R. Event-related potentials and the recollection of associative
information. Cogn. Brain. Res., 1996, 4: 297–304.
Salmaso, D. and Longoni, A.M. Problems in the assessment of hand
preference. Cortex, 1985, 21: 533–549.
Scheffers, M.K., Johnson, R. Jr. and Ruchkin, D.S. P300 in patients with
unilateral temporal lobectomies: the effects of reduced stimulus quality.
Psychophysiology, 1991, 28: 274–284.
Sharbrough, F., Chatrian, G.E., Lesser, R.P., Lüders, H., Nuwer, M. and
Picton, T.W. American Electroencephalographic Society guidelines for
standard electrode position nomenclature. J. Clin. Neurophysiol., 1991,
8: 200–202.
Smith, M.E., Stapleton, J.M. and Halgren, E. Human medial temporal lobe
potentials evoked in memory and language tasks. Electroenceph. clin.
Neurophysiol., 1986, 63: 145–159.
So, N.K. Depth electrode studies in mesial temporal lobe epilepsy. In: H.
Lüders (Ed.), Epilepsy Surgery. Raven Press, New York, 1991, pp.
371–384.
Verleger, R., Lefebre, C., Wieschemeyer, R. and Kompf, D. Event related
potentials suggest slowing of brain processes in generalized epilepsy
and alterations of visual processing in patients with partial seizures.
Cogn. Brain Res., 1997, 5: 205–219.
Wyler, A.R. Subdural strip electrodes in surgery of epilepsy. In: H.
Lüders (Ed.), Epilepsy Surgery. Raven Press, New York, 1991, pp.
371–384.