- Email: [email protected]
Epilepsy increases vulnerability of long-term face recognition to proactive interference
Epilepsy & Behavior 8 (2006) 220–227 www.elsevier.com/locate/yebeh
Epilepsy increases vulnerability of long-term face recognition to proactive interference T. Bengner a,*, T. Malina a, M. Lindenau a, B. Voges a, E. Goebell b, S. Stodieck a a
b
Epilepsy Center Hamburg, Protestant Hospital Alsterdorf, Hamburg, Germany Department of Neuroradiology, University Clinic Eppendorf, University of Hamburg, Hamburg, Germany Received 20 September 2005; revised 13 October 2005; accepted 14 October 2005 Available online 13 December 2005
Abstract Proactive interference (PI) decreases short- and long-term memory in healthy subjects. Neurological patients exhibit a heightened PI effect on short-term memory. It is, however, not known if PI affects long-term memory in neurological patients. We analyzed whether epilepsy heightens the negative effect of PI on long-term face memory. PI was induced by a list of 20 faces learned 24 hours prior to a target list of 20 faces. We tested immediate and 24-hour recognition for both lists. Twelve healthy controls and 42 patients with generalized epilepsy or temporal lobe epilepsy (TLE) were studied. PI led to a decrease in 24-hour recognition in patients with generalized epilepsy and TLE but not in controls. Thus, PI may cause long-term memory disturbances in epilepsy patients. PI was also associated with decreased short-term memory, but only in right TLE. This confirms the dominant role of the right temporal lobe in short-term face memory. 2005 Elsevier Inc. All rights reserved. Keywords: Temporal lobe epilepsy; Generalized epilepsy; Face memory; Short-term memory; Forgetting; MRI-negative; Nonlesional epilepsy; Interictal epileptiform abnormalities; Delayed recognition; Immediate recognition; Cognition; Neuropsychology; Hemispheric dominance
1. Introduction In proactive interference (PI), previously learned information impairs learning or remembrance of more recent material. For example, forgetting the parking space where you most recently left the car in a parking lot might be influenced by the memory of former stops at this lot. Experimentally, PI is usually induced by one or more distracter lists of information that are learned before the acquisition of a similar target list. Traditionally, these are paired-associate or serial lists. The distracter lists can be presented all at once right before the target list (massed PI) or distributed over days (distributed PI). It was shown that PI decreases short-term memory [1,2] and 24-hour long-term memory (for review, see [3]). However, the detrimental effect of PI on long-term memory in healthy indi*
Corresponding author. Fax: +49 0 40 5077 4942. E-mail address: [email protected] (T. Bengner).
1525-5050/$ - see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.yebeh.2005.10.009
viduals is visible only with massed, but not with distributed, PI [4]. The latter point contributed to the abandonment of PI as a major source of forgetting in healthy individuals, presupposing that most PI in everyday life occurs in a distributed manner (for review, see [5]). It is nevertheless unknown whether distributed PI may still decrease long-term memory in neurological patients. In this study we tested the hypothesis that distributed PI has a more pronounced influence on long-term face recognition in epilepsy patients than in healthy controls. The present study furthermore addressed the effect of PI on short-term face recognition in temporal lobe epilepsy (TLE). PI decreases short-term face memory in healthy subjects [6], and there is also evidence that PI has a greater negative effect on short-term memory in neurological patients when compared with healthy controls [7–10]. It is known that the left temporal lobe is specialized in verbal memory, while the right temporal lobe is dominant for nonverbal memory [11–13]. Consistent with this, the influ-
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
221
mainly by the epileptogenic focus rather than the morphological lesion itself [22]. Previous studies have focused primarily on patients with lesional TLE or mixed groups of lesional and MRI-negative patients (e.g., [23,24]). Thus, further investigation of the effects of MRI-negative epilepsy in isolation is required to determine how memory is influenced by more subtle brain alterations related to epilepsy that are not detected by current MRI. The present study investigated only patients with MRI-negative generalized epilepsy and TLE. In summary, we analyzed the influence of distributed PI on short- and long-term face recognition in patients with MRI-negative epilepsy, comprising TLE, generalized epilepsy, and healthy controls. PI was induced by a list of 20 faces learned 24 hours prior to a target list of 20 faces (see Fig. 1). We tested immediate and 24-hour recognition of both lists. Thus, subjects had already looked at 100 photographs of 60 different faces before learning the target list. 2. Methods 2.1. Subjects
Fig. 1. Illustration of the study design. Proactive interference was induced by a distracter list of 20 unfamiliar faces that was learned 24 hours before a similar target list of again 20 unfamiliar faces. After the distracter list was learned, immediate and 24-hour recognition was tested. For recognition testing, 20 unknown faces were presented together with the learned faces. Immediately after the test of 24-hour recognition of the distracter list, subjects learned the target list, a new list of 20 unfamiliar faces, and they were then tested again for recognition in the same manner immediately and after 24 hours. This ‘‘distributed’’ PI paradigm differs from ‘‘massed’’ PI paradigms by stretching the presentation of the distracter information over days rather than inducing PI by presenting distracter information only immediately before the presentation of the target information. For further details, please refer to Section 2.
ence of PI on short-term memory in TLE is material specific, depending on the lateralization of the lesion. Patients with left TLE (LTLE) but not right TLE (RTLE) exhibit a detrimental effect of PI on verbal short-term memory [14–16]. Based on the dominance of the right temporal lobe in short-term face memory [17–21], we hypothesized that PI should have a detrimental effect on short-term face memory in RTLE but not in LTLE. People with TLE are a heterogeneous group, including patients with hippocampal sclerosis or other morphological alterations and lesions, as well as those with normal MRI findings. Memory disturbances in TLE may be caused
Subjects were 12 volunteers and 42 patients of the Epilepsy Center Hamburg. All participants gave informed consent to the study. All patients were undergoing a diagnostic workup with 48- to 72-hour continuous video/EEG monitoring and structural MRI investigations between April 2004 and May 2005. Dosage of medication was not lowered during video/EEG monitoring, but some of the patients were in a transition phase between different antiepileptic medication regimens. As the patients were not candidates for epilepsy surgery, reduction of medication was not necessary. Epilepsy was diagnosed when recurrent unprovoked seizures of an unquestionable epileptic nature could be verified from the patientÕs history or during video/ EEG monitoring. The diagnosis of TLE was based on reliable information about clinical seizure patterns typical of a temporal focus (aura and ictal clinical phenomena). The necessary information on seizure history and semiology was taken from interviews with the patient or witnesses of the seizures or from our own video material. During video/EEG monitoring, 11 patients exhibited left temporal interictal epileptic abnormalities (IEAs, LTLE), 9 patients exhibited right temporal IEAs (RTLE), and 10 patients had TLE with no IEAs (TLE, no IEAs) (for details on video/EEG monitoring, see the next section). Please note that patients with TLE were assigned to the RTLE or LTLE group solely on the basis of their IEAs, as most patients did not have any seizures during video/EEG monitoring (see Table 1). This method was considered valid considering the fact that other studies had found unilateral IEAs to be a good lateralizing feature of the epileptogenic region in MRI-negative TLE [25,26]. Patients without IEAs were diagnosed as having TLE only if a sufficient number of auras or seizures could be classified as typical of a temporal focus. In addition, we included 12 patients with idiopathic
222
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
Table 1 Characteristics of patients and controls
Age, mean (SD) Verbal IQ (vocabulary), mean (SD) Sex (female/male) Mean duration of epilepsy (years) Handedness (right/left) Seizures, first 24-hour delay (no/yes) Seizures, second 24-hour delay (no/yes) Psychiatric comorbidity (no/yes) IEA frequency (rare/intermittent/frequent) Sleep deprivation, first 24-hour delay Sleep deprivation, second 24-hour delay
LTLE (N = 11)
TLE, no IEA (N = 10)
RTLE (N = 9)
Generalized epilepsy (N = 12)
Controls (N = 12)
P
45 (16) 99 (8) 9/2 18 8/3 11/0 11/0 7/4 3/8/0 1 4
48 (14) 101 (9) 8/2 7 9/1 10/0 10/0 5/5 — 2 3
37 (7) 100 (8) 4/5 14 9/0 7/2 8/1 4/5 2/4/3 3 0
29 (11) 98 (7) 10/2 13 12/0 10/2 12/0 7/5 — 0 1
28 (11) 100 (11) 7/5 — 12/0 — — — — — —
<0.001a 0.90a 0.21b 0.15c 0.06b 0.14b 0.28b 0.83b 0.11b 0.08b 0.12b
LTLE, left temporal lobe epilepsy; RTLE, right temporal lobe epilepsy; IEA, interictal epileptiform abnormalities; TLE, no IEA, temporal lobe epilepsy with no IEAs detected during video-EEG monitoring; IQ, intelligence quotient. a ANOVA. b 2 v test. c Kruskal–Wallis ANOVA.
generalized epilepsy. Idiopathic generalized epilepsy was defined by typical semiology of juvenile myoclonic epilepsy, absence seizures, or primary generalized tonic–clonic seizures and generalized EEG patterns during video/EEG monitoring. Detailed structural MRI investigations were normal in all patients (for MRI details, see Section 2.3). The healthy control subjects were employees from different departments of our clinic. Among all patients, 5 had seizures and 14 were sleep-deprived for diagnostic reasons during the 24-hour delays between the two test phases (see Table 1). Seizures in TLE patients were complex partial seizures and showed ictal EEG abnormalities in accordance with the hemisphere in which IEAs had been detected. Patients with generalized epilepsy had primary generalized seizures and absence seizures. The psychiatric status of the patients was determined by patient history and clinical interviews. If necessary, questionnaires, the Beck Depression Inventory and the State–Trait–Anxiety Inventory, were used. Nineteen patients (45% of patients) received a psychiatric diagnosis in addition to epilepsy. Comorbidities included anxiety disorder (N = 6), depression (N = 10), mild psychosis (N = 1), insecure personality disorder (N = 1), and nonepileptic seizures (N = 1). 2.2. Noninvasive continuous video/EEG monitoring We employed 32- to 64-channel EEG. Electrodes were placed according to the 10/20 system, with additional electrodes according to the 10/10 system. EEG data were continuously checked for IEAs and seizure patterns by trained assistant medical technicians and evaluated by authors M.L., B.V., or S.S. Only definitive spike–slow waves were rated as IEAs. In those with TLE, IEAs were localized to electrodes F7/8, T1/2, T3/4, and T5/6. Background abnormalities or pathological slow waves were ignored for the purpose of this study. Spike–slow wave frequency was clinically rated as ‘‘none,’’ ‘‘rare,’’ ‘‘intermittent,’’ or ÔfrequentÕ
in patients with TLE (see Table 1). In patients with generalized epilepsy, generalized epileptic activity occurred mainly under hyperventilation or during sleep, and was very rare in 10 cases and occurred intermittently in 2 cases. As epileptic activity in generalized epilepsy is not directly comparable to IEAs in TLE, it was not included in Table 1. IEAs did not have to be evident during the 24-hour delays, although they were in cases with intermittent and frequent IEAs and in most of the patients with rare IEAs. However, patients were included in the left or right groups when they exhibited IEAs at any time during their stay in the EEG unit. Sleep-deprived patients had the opportunity to partly compensate for their sleep deficit before the 24-hour recognition. 2.3. Structural MRI investigations MRI scans were acquired with a 1.5-T scanner (Siemens Magnetom Symphony, Erlangen/Germany). Imaging studies included transverse T2-weighted turbo spin echo/proton density images, transverse FLAIR images, coronal T2weighted images (covering the hippocampus in a plane perpendicular to its long axis), transverse T1-weighted images, a sagittal T1-weighted MPRage three-dimensional volume scan, and transverse T1-weighted images postadministration of contrast medium. MRI studies of patients performed at other centers were reevaluated by the authors B.V., M.L., S.S., and E.G. and presented at an interdisciplinary neuroradiological case conference. In case the external MRIs did not meet the standards of the imaging protocol outlined earlier, patients were restudied at our center. 2.4. Experimental design We used 662 black-and-white face photographs (355 male), taken from a front perspective (age, 25–35). Hair was visible, but no glasses, beards, or excessive jewels.
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
Subjects received an individual set of learning and test phase items to avoid possible conversations between patients about face item characteristics. A computer program by author T.M. controlled the random draft and presentation scheme of faces and saved time of learning and test phases and the subjectsÕ answers and reaction times. During their stay in the video/EEG monitoring unit, patients learned a series of 20 unfamiliar faces presented on day 1 (distracter list); this was followed by an immediate recognition test, in which the learned faces had to be detected among 20 old and 20 new faces. After 24 hours, on day 2, recognition was tested again. After that, 20 new unfamiliar faces had to be learned (target list). Patients were tested for recognition of this second list immediately and again after 24 hours on day 3 (see Fig. 1). A 17-in. screen was employed, and photographs were 19 · 28 cm. Healthy controls were tested with a 15-in. screen, leading to a slightly smaller size of the faces presented. Before the learning phase started, subjects were verbally instructed to carefully look at the faces, as they would have to recognize them from a greater number of faces later. Twenty faces randomly picked from the face item pool were presented one at a time for 5 seconds on a computer screen. During presentation, ‘‘learning phase’’ (Lernphase) was written on the upper left corner of the screen. The first test occurred after the learning phase. An array of 40 faces in the same format were shown serially; this contained the 20 old faces randomly mixed with 20 new faces that were again randomly picked from the remaining face item pool. Subjects were instructed to decide whether they had seen the faces during the learning phase or not. Patients were informed that a second test would take place the next day. Twenty-four hours later, 40 faces (20 old faces shuffled with 20 new faces randomly taken from the face item pool) were again presented to the subjects. Subjects were instructed to decide whether they had seen the face during the learning phase the day before or not. Patients received feedback about the number of correct hits and correct rejections immediately after each test. After the 24-hour recognition test, subjects learned a new list of 20 unfamiliar faces, again randomly picked from the face item pool. Before the second learning phase started, they were informed that a new identical test would begin now, and that all faces seen so far would not be repeated anymore in the tests to come. The same round comprising learning, immediate recognition, and 24-hour delayed recognition was repeated as described earlier.
223
2.5. Statistical methods The main dependent variable was the corrected percentage of recognized faces, calculated as (hits false positives) · 100/20. We further evaluated true positive and true negative reactions and the reaction times for true positives, true negatives, false positives, and false negatives. For statistical analysis, repeated-measures analysis of covariance (RANCOVA), repeated-measures analysis of variance (RANOVA), analysis of variance (ANOVA), Kruskal–Wallis ANOVA, v2 tests, Mann–Whitney tests, and Wilcoxon tests were used. When group sizes were too small for parametric analysis (RANCOVA), we reanalyzed data with nonparametric statistics. As both methods lead to equal results and RANCOVA is more succinct in describing interaction effects, we report only these results. Significant results were further explored by specified contrasts and Scheffe´ tests. 3. Results Table 1 lists differences in clinical history and general data between the groups. Table 2 summarizes the mean face recognition performance (corrected for false positives) of the different groups. A 5 · 2 · 2 RANCOVA with the factors group (LTLE/RTLE/TLE, no IEAs/generalized epilepsy/controls), time point (immediate vs 24-hour recognition), and list (distracter list vs target list) was calculated. The covariates age, handedness, sex, sleep deprivation, and seizures during the two delays were included. As none of the covariates had a moderating influence, we report results without the inclusion of covariates. For an illustration of the results, see Fig. 2. 1. We found a significant main effect for the group factor (F = 3.9; P = 0.008). Post hoc Scheffe´ analysis revealed overall worse face recognition in patients with RTLE than in healthy controls (P = 0.01). No other groups differed from each other using post hoc analysis (P > 0.24). 2. Face recognition became worse for the target list (F = 6.9; P = 0.01). 3. Twenty-four hour recognition was worse than immediate recognition (F = 17.8; P < 0.001). 4. There was an interaction between group and time point (F = 4.2; P = 0.005). Contrast analysis revealed that healthy controls and patients with generalized epilepsy did not differ between immediate and 24-hour recognition (P > 0.32), whereas all TLE groups had worse 24-
Table 2 Mean face recognition of the different groupsa
Distractor Target a
Immediate 24-hour Immediate 24-hour
Controls
RTLE
LTLE
TLE, no IEA
Generalized epilesy
75 76 79 73
64 45 52 39
64 60 70 47
66 62 65 49
63 78 63 55
(16) (13) (15) (17)
Corrected for false positives, SD in parentheses.
(14) (13) (14) (17)
(26) (25) (16) (34)
(14) (11) (19) (23)
(19) (18) (17) (32)
224
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
Fig. 2. Face recognition of distracter and target lists. Values are means; bars denote 0.95 confidence intervals. For further details, please refer to Section 3.
hour recognition than immediate recognition (P < 0.02). In comparison, patients with generalized epilepsy showed even better 24-hour recognition than immediate recognition of the distracter face list (P < 0.001). Their 24-hour recognition of the distracter list did not differ from that of healthy controls (P = 0.74). In comparison, patients with TLE with right, left, or no IEAs were worse than healthy controls in their 24-hour recognition of the distracter list (P < 0.001, P = 0.04, and P = 0.06, respectively). 5. We found an interaction between list and time point (F = 9.3; P = 0.004). Contrast analysis showed that immediate and delayed recognition did not differ for the distracter list (P > 0.34), whereas delayed recognition was worse than immediate recognition for the target list (P < 0.001, compare Fig. 2). The latter effect was explained as follows: immediate recognition did not differ between the distracter and target lists (P = 0.92), whereas delayed recognition was worse for the target list than for the distracter list (P < 0.001; compare Fig. 2). As a number of patients had a psychiatric comorbidity, we wanted to exclude the possibility that the interaction of list and time point was caused by the psychiatric comorbidity rather than epilepsy. Thus, we tested whether the effect would still hold if we excluded patients with a psychiatric comorbidity and reanalyzed the data accordingly. The number of patients decreased to a total of 23 (for details, see Table 1). The interaction between time point and list was still visible (N = 35, P = 0.04). It is thus fairly unlikely that the psychiatric diagnosis in a number of the patients could explain the found effect. 6. The three-way interaction failed to reach significance (F = 1.8; P = 0.14). However, while patients with LTLE and generalized epilepsy did show a steeper decrease
from immediate to 24-hour recognition for the target than the distracter list (contrast analyses, P < 0.001 and P = 0.02, respectively), healthy controls, patients with TLE without IEAs, and patients with RTLE did not exhibit this effect (contrast analyses, P > 0.16). Patients with TLE without IEAs still showed a significant decrease from immediate to delayed recognition on the target list (contrast analysis, P = 0.03), but not on the distracter list (contrast analysis, P = 0.38). The RTLE group showed a decrease in immediate recognition from the first to the second list (P = 0.05), unlike all other groups (contrast analysis, P = 0.03). The RTLE group also differed from all other groups in showing a smaller decrease from immediate to delayed recognition on the target than on the distracter list (contrast analysis, P = 0.04), due to the decrease in immediate recognition of the target list. As a number of patients were sleep-deprived or had seizures during the 24-hour delays, we wanted to make sure that the interaction of list and time point was caused by epilepsy rather than these factors. We thus tested whether the effect would still hold if we excluded patients who were sleep-deprived or had seizures and reanalyzed the data accordingly. The number of patients decreased to a total of 25 (for details, see Table 1; note that one sleep-deprived patient had a seizure, and another patient had a seizure during the first and second delays). Overall, the interaction between time point and list was still apparent; i.e., longterm recognition of the target list was still worse than long-term recognition of the distracter list (N = 37, P < 0.05). The healthy controls did not show this interaction effect (contrast analysis, P = 0.39), whereas the patients with LTLE, TLE without IEAs, and generalized
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
epilepsy (N = 20) still showed the interaction effect (contrast analysis, P < 0.05). It is thus fairly unlikely that sleep deprivation or seizures in a number of patients could explain the found effect. Due to the exclusion of patients with sleep deprivation and seizures, the different patient groups were too small for further separate analysis. Although age had no influence on the results as a covariate, it might be that epilepsy patients would not differ from healthy controls were they age matched. To test this interpretation, we excluded patients aged above 55 or below 22. This led to a nonsignificant difference in age between the different groups (N = 39; F = 0.62; P = 0.65). Post hoc Scheffe´ tests were all nonsignificant as well (PÕs > 0.76). The different group sizes were still large enough for further statistical evaluation (controls: N = 7; LTLE, N = 8, RTLE: N = 9; TLE without IEAs: N = 7; generalized epilepsy: N = 8). The RANOVA was recalculated and still showed the interaction between list and time point (F = 6.4; P = 0.02). Although healthy controls did not show this interaction (N = 7; contrast analysis: P = 0.47), the patients still exhibited this effect (N = 32; contrast analysis: P = 0.02). Consistent with the initial results, this significant interaction was due mainly to the LTLE (contrast analysis: P = 0.07) and generalized epilepsy (contrast analysis: P = 0.01) groups. In summary, it is fairly unlikely that age had a moderating influence on the difference between controls and epilepsy patients in the found interaction between time point and list. To better understand the decreased delayed recognition of the target list, we reanalyzed our data by splitting the dependent variable into correctly recognized learned faces (true positives) and correctly rejected new faces (true negatives). A 5 · 2 · 2 · 2 RANOVA with the factors group (TLE left IEAs/TLE right IEAs/TLE no IEAs/generalized
225
epilepsy/controls), time point, list, and recognition type (true positives vs true negatives) was calculated. For an illustration of the results see Fig. 3. Apart from the effects already described, we found an interaction between list, time point, and recognition type (F = 6.3; P = 0.02). Contrast analysis revealed that while the number of true positives decreased from the distracter to the target list, whether immediate or 24-hour recognition was being tested (P = 0.02), the number of true negatives differed between the distracter and target lists, depending on whether it was immediate or 24-hour recognition. Although the number of true negatives during immediate recognition was higher for the target than the distracter list (contrast analysis, P = 0.02), it was lower for the target than the distracter list during 24-hour recognition (contrast analysis, P < 0.001) (see Fig. 2). While the number of true negatives from the distracter list did not differ between immediate and 24-hour recognition (P = 0.31), fewer true negatives were visible during delayed than immediate recognition (P < 0.001). No three-way interaction was visible (P = 0.73). In summary, the detrimental effect of PI on 24-hour face recognition is due mainly to a combined decrease of true positive and true negative reactions during 24-hour recognition. We also analyzed reaction times for true positive, true negative, false positive, and false negative responses. A 5 · 2 · 2 · 4 RANOVA with the factors group (TLE left IEAs/TLE right IEAs/TLE no IEAs/generalized epilepsy/ controls), time point, list, and reaction time (true positive, true negative, false positive, and false negative responses) was calculated. Mean reaction times increased during delayed recognition trials (immediate: 3.2 seconds, SE = 0.18; 24-hour: 3.6 seconds, SE = 0.21; F = 5.0; P = 0.03). False responses took longer than correct
Fig. 3. True positive and true negative reactions during immediate and delayed recognition of the distracter and target lists. Values are means; bars denote 0.95 confidence intervals. For further details, please refer to the Section 3.
226
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
responses (mean: false positives: 4.3 seconds, SE = 0.37; false negatives: 3.7 seconds, SE = 0.25; true positives: 2.6 seconds, SE = 0.10; true negatives: 3.0 seconds, SE = 0.13; F = 31.5; P < 0.001). False negative and false positive responses did not differ (F = 2.5; P = 0.12), but true positive responses were faster than true negative responses (F = 13.0; P < 0.001). No other effects were detected (P P 0.12). To better understand the face recognition deficit in RTLE, we analyzed the correlations between 24-hour recognition of the distracter list and that of the target list. The correlations were as follows: healthy controls, r = 0.82 (P = 0.01); RTLE, r = 0.25 (P = 0.51); LTLE, r = 0.65 (P = 0.03); TLE, no IEAs, r = 0.34 (P = 0.34); generalized epilepsy, r = 0.27 (P = 0.47). The correlations between RTLE and healthy controls differed (P = 0.02, two-tailed). No other groups differed from each other in this respect. 4. Discussion The present study analyzed the influence of distributed PI on short- and long-term face memory in patients with MRI-negative TLE, patients with generalized epilepsy, and healthy controls. The main results of the study are that (1) PI led to decrease in 24-hour delayed recognition, but not to a decline in immediate recognition; (2) this effect was significant in patients with LTLE, patients with generalized epilepsy, and, to a lesser degree, in patients with TLE without IEAs (in comparison, it was not significant in healthy controls and patients with RTLE); and (3) in opposition to the other groups, PI led to decreased short-term memory in patients with RTLE. To our knowledge, this study demonstrates for the first time a heightened effect of PI on long-term recognition in epilepsy patients. Our results suggest that although a distributed PI paradigm does not decrease 24-hour recognition in healthy subjects [4], distributed PI may be a source of long-term memory disturbance in epilepsy patients. The results complement studies that have found an influence of massed PI on 24-hour long-term memory in healthy individuals [3,4]. The mechanisms underlying PI are not known. However, evidence has been found for a role of the prefrontal cortex and frontopolar cortex in resolving the effect of PI in short-term recognition [27–29]. In our study, LTLE and generalized epilepsy were related to a stronger effect of PI on memory than TLE without IEAs or healthy controls, suggesting that prefrontal lobe function is negatively influenced in LTLE or generalized epilepsy. This conjecture is supported by studies that found frontal and prefrontal dysfunction in LTLE and generalized epilepsy [e.g., [30–33]]. Our result that PI affects 24-hour recognition but not immediate recognition is in agreement with studies in healthy subjects, in whom the effect of PI on memory was observed to increase with the passage of time between learning and recall [e.g., [34]].
The present study also shows for the first time that PI affects immediate face recognition in RTLE but not LTLE. This finding complements studies that reported a detrimental effect of PI on verbal short-term memory in LTLE [14– 16]. Interestingly, PI did not lead to a further decrease in 24-hour delayed recognition in patients with RTLE, as was visible in LTLE, generalized epilepsy, and TLE without IEAs. Perhaps the distracter list causes less disturbance in the recognition of faces from the target list 24 hours later, because patients with RTLE already exhibited a deficit in recognition of the distracter list. The significantly lower and negative correlation between 24-hour recognition of the distracter list and that of the target list in RTLE compared with the high positive correlation in healthy controls supports this interpretation. However, the present study could not determine whether PI has an effect on long-term face recognition in RTLE. This can only be investigated if performance in the recognition of the distracter list is matched in all groups, as was done in studies of healthy controls [3]. Rather, our study suggests that it may be more difficult to further induce a detrimental effect of PI on the already decreased long-term face memory in RTLE. That patients with RTLE had an overall worse face memory compared with healthy controls confirms the dominant role of the right temporal lobe in face memory [17–21]. In the present study, PI was elicited by learning a list of 20 faces 24 hours before learning a second list of 20 faces, in accordance with a distributed PI paradigm (compare [4]). Immediately before learning the target list, subjects were again presented with 40 faces for testing 24-hour recognition of the distracter list. It is hence not possible to distinguish whether the PI effect on long-term memory evident in this study was caused by learning the distracter list, or by its presentation during the 24-hour recognition prior to learning of the target list, or by a combination of both. However, the aim of the present study was to establish whether epilepsy heightens the effect of PI on long-term memory. Future studies can determine under what conditions this is the case. The influence of PI on memory as seen in the patients studied might be a nonspecific effect of antiepileptic medication, which was not controlled for in this study. A recent study found accelerated forgetting in patients with high serum levels of antiepileptic medications [35]. Another study found that differences between LTLE and RTLE with respect to the effect of proactive and retroactive interference on verbal short-term memory were visible only under full medication, but not under moderately reduced anticonvulsant medication [14]. However, as most epilepsy patients usually depend on anticonvulsant medication, the present study at least proves that distributed PI has a specific impact on short-term and long-term face memory in epilepsy patients under their regular medication regimen. Our study was restricted to patients with MRI-negative epilepsy. Although this restriction raised the homogeneity of the studied sample and made the comparison of patients with TLE, patients with generalized epilepsy, and healthy
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
controls easier, it is not known whether lesional epilepsy patients will exhibit the same susceptibility to PI. A majority of patients with TLE show hippocampal sclerosis or other lesions. Thus it is up to future studies to prove whether PI decreases short- and long-term face recognition in MRI-positive, lesional TLE. Based on our observation that PI impairs memory in epilepsy patients, the results of this study might be clinically relevant. For example, face memory tests eliciting PI in combination with longer testing delays between learning and recognition could increase the sensitivity of detecting memory disturbances in epilepsy, e.g., when patients complain about memory deficits in everyday life but perform normally on standard memory tests. References [1] Keppel G, Underwood BL. Proactive inhibition in short-term retention of single items. J Verb Learn Verb Behav 1962;1:153–61. [2] Shimamura AP, Jurica PJ, Mangels JA, Gershberg FB. Susceptibility to memory interference effects following frontal lobe damage: findings from tests of paired-associate learning. J Cogn Neurosci 1995;7:144–52. [3] Underwood BL. Interference and forgetting. Psychol Rev 1957;64:49–60. [4] Underwood BJ, Ekstrand BR. An analysis of some shortcomings in the interference theory of forgetting. Psychol Rev 1966;73:540–9. [5] Wixted JT. The psychology and neuroscience of forgetting. Annu Rev Psychol 2004;55:235–69. [6] Yarmey AD. Proactive interference in short-term retention of human faces. Can J Psychol 1974;28:333–8. [7] Flicker C, Ferris SH, Crook T, Bartus RT. Age differences in the vulnerability of facial recognition memory to proactive interference. Exp Aging Res 1989;15:189–94. [8] Warrington EK, Weiskrantz L. The effect of prior learning on subsequent retention in amnesic patients. Neuropsychologia 1974;12:419–28. [9] Warrington EK, Weiskrantz L. Further analysis of the prior learning effect in amnesic patients. Neuropsychologia 1978;16:169–77. [10] Moscovitch M, Winocur G. Contextual cues and release from proactive inhibition in young and old people. Can J Psychol 1983;37:331–44. [11] Milner B. Psychological aspects of focal epilepsy and its neurosurgical management. In: Purpura D, Penry J, Walter R, editors. Advances in neurology: vol. 8. Neurosurgical management of the epilepsies. New York: Raven Press; 1975. p. 299–321. [12] Ladavas E, Umilta C, Provinciali L. Hemisphere-dependent cognitive performances in epileptic patients. Epilepsia 1979;20:493–502. [13] Morris RG, Polkey CE, Cox T. Recognition memory for words and faces following unilateral temporal lobectomy. Br J Clin Psychol 1995;34:571–6. [14] Durwen HF, Elger CE. Verbal learning differences in epileptic patients with left and right temporal lobe foci: a pharmacologically induced phenomenon? Acta Neurol Scand 1993;87:1–8. [15] Giovagnoli AR, Avanzini G. Forgetting rate and interference effects on a verbal memory distractor task in patients with temporal lobe epilepsy. J Clin Exp Neuropsychol 1996;18:259–64.
227
[16] Winocur G, Moscovitch M, Bruni J. Heightened interference on implicit, but not explicit, tests of negative transfer: evidence from patients with unilateral temporal lobe lesions and normal old people. Brain Cogn 1996;30:44–58. [17] Milner B. Visual recognition and recall after right temporal-lobe excision in man. Neuropsychologia 1968;6:191–209. [18] Barr WB. Examining the right temporal lobeÕs role in nonverbal memory. Brain Cogn 1997;35:26–41. [19] Kelley WM, Miezin FM, McDermott KB, et al. Hemispheric specialization in human dorsal frontal cortex and medial temporal lobe for verbal and nonverbal memory encoding. Neuron 1998;20:927–36. [20] Kelley WM, Ojemann JG, Wetzel RD, et al. Wada testing reveals frontal lateralization for the memorization of words and faces. J Cogn Neurosci 2002;14:116–25. [21] Coleshill SG, Binnie CD, Morris RG, et al. Material-specific recognition memory deficits elicited by unilateral hippocampal electrical stimulation. J Neurosci 2004;24:1612–6. [22] Giovagnoli AR, Avanzini G. Learning and memory impairment in patients with temporal lobe epilepsy: relation to the presence, type, and location of brain lesion. Epilepsia 1999;40:904–11. [23] Baxendale SA. The role of the hippocampus in recognition memory. Neuropsychologia 1997;35:591–8. [24] Glogau S, Ellgring H, Elger CE, Helmstaedter C. Face and facial expression memory in temporal lobe epilepsy patients: preliminary results. Epilepsy Behav 2004;5:106–12. [25] Holmes MD, Born DE, Kutsy RL, Wilensky AJ, Ojemann GA, Ojemann LM. Outcome after surgery in patients with refractory temporal lobe epilepsy and normal MRI. Seizure 2000;9:407–11. [26] Sylaja PN, Radhakrishnan K, Kesavadas C, Sarma PS. Seizure outcome after anterior temporal lobectomy and its predictors in patients with apparent temporal lobe epilepsy and normal MRI. Epilepsia 2004;45:803–8. [27] Jonides J, Marshuetz C, Smith EE, Reuter-Lorenz PA, Koeppe RA, Hartley A. Age differences in behavior and PET activation reveal differences in interference resolution in verbal working memory. J Cogn Neurosci 2000;12:188–96. [28] Badre D, Wagner AD. Frontal lobe mechanisms that resolve proactive interference. Cereb Cortex 2005;15:2003–12. [29] King JA, Hartley T, Spiers HJ, Maguire EA, Burgess N. Anterior prefrontal involvement in episodic retrieval reflects contextual interference. Neuroimage 2005;28:256–67. [30] Jokeit H, Seitz RJ, Markowitsch HJ, Neumann N, Witte OW, Ebner A. Prefrontal asymmetric interictal glucose hypometabolism and cognitive impairment in patients with temporal lobe epilepsy. Brain 1997;120:2283–94. [31] Swartz BE, Simpkins F, Halgren E, et al. Visual working memory in primary generalized epilepsy: an 18FDG-PET study. Neurology 1996;47:1203–12. [32] Holmes MD, Brown M, Tucker DM. Are ‘‘generalized’’ seizures truly generalized? Evidence of localized mesial frontal and frontopolar discharges in absence. Epilepsia 2004;45:1568–79. [33] Zifkin B, Andermann E, Andermann F. Mechanisms, genetics, and pathogenesis of juvenile myoclonic epilepsy. Curr Opin Neurol 2005;18:147–53. [34] Briggs GE. Acquisition, extinction, and recovery functions in retroactive inhibition. J Exp Psychol 1954;47:285–93. [35] Jokeit H, Kramer G, Ebner A. Do antiepileptic drugs accelerate forgetting? Epilepsy Behav 2005;6:430–2.
Epilepsy increases vulnerability of long-term face recognition to proactive interference T. Bengner a,*, T. Malina a, M. Lindenau a, B. Voges a, E. Goebell b, S. Stodieck a a
b
Epilepsy Center Hamburg, Protestant Hospital Alsterdorf, Hamburg, Germany Department of Neuroradiology, University Clinic Eppendorf, University of Hamburg, Hamburg, Germany Received 20 September 2005; revised 13 October 2005; accepted 14 October 2005 Available online 13 December 2005
Abstract Proactive interference (PI) decreases short- and long-term memory in healthy subjects. Neurological patients exhibit a heightened PI effect on short-term memory. It is, however, not known if PI affects long-term memory in neurological patients. We analyzed whether epilepsy heightens the negative effect of PI on long-term face memory. PI was induced by a list of 20 faces learned 24 hours prior to a target list of 20 faces. We tested immediate and 24-hour recognition for both lists. Twelve healthy controls and 42 patients with generalized epilepsy or temporal lobe epilepsy (TLE) were studied. PI led to a decrease in 24-hour recognition in patients with generalized epilepsy and TLE but not in controls. Thus, PI may cause long-term memory disturbances in epilepsy patients. PI was also associated with decreased short-term memory, but only in right TLE. This confirms the dominant role of the right temporal lobe in short-term face memory. 2005 Elsevier Inc. All rights reserved. Keywords: Temporal lobe epilepsy; Generalized epilepsy; Face memory; Short-term memory; Forgetting; MRI-negative; Nonlesional epilepsy; Interictal epileptiform abnormalities; Delayed recognition; Immediate recognition; Cognition; Neuropsychology; Hemispheric dominance
1. Introduction In proactive interference (PI), previously learned information impairs learning or remembrance of more recent material. For example, forgetting the parking space where you most recently left the car in a parking lot might be influenced by the memory of former stops at this lot. Experimentally, PI is usually induced by one or more distracter lists of information that are learned before the acquisition of a similar target list. Traditionally, these are paired-associate or serial lists. The distracter lists can be presented all at once right before the target list (massed PI) or distributed over days (distributed PI). It was shown that PI decreases short-term memory [1,2] and 24-hour long-term memory (for review, see [3]). However, the detrimental effect of PI on long-term memory in healthy indi*
Corresponding author. Fax: +49 0 40 5077 4942. E-mail address: [email protected] (T. Bengner).
1525-5050/$ - see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.yebeh.2005.10.009
viduals is visible only with massed, but not with distributed, PI [4]. The latter point contributed to the abandonment of PI as a major source of forgetting in healthy individuals, presupposing that most PI in everyday life occurs in a distributed manner (for review, see [5]). It is nevertheless unknown whether distributed PI may still decrease long-term memory in neurological patients. In this study we tested the hypothesis that distributed PI has a more pronounced influence on long-term face recognition in epilepsy patients than in healthy controls. The present study furthermore addressed the effect of PI on short-term face recognition in temporal lobe epilepsy (TLE). PI decreases short-term face memory in healthy subjects [6], and there is also evidence that PI has a greater negative effect on short-term memory in neurological patients when compared with healthy controls [7–10]. It is known that the left temporal lobe is specialized in verbal memory, while the right temporal lobe is dominant for nonverbal memory [11–13]. Consistent with this, the influ-
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
221
mainly by the epileptogenic focus rather than the morphological lesion itself [22]. Previous studies have focused primarily on patients with lesional TLE or mixed groups of lesional and MRI-negative patients (e.g., [23,24]). Thus, further investigation of the effects of MRI-negative epilepsy in isolation is required to determine how memory is influenced by more subtle brain alterations related to epilepsy that are not detected by current MRI. The present study investigated only patients with MRI-negative generalized epilepsy and TLE. In summary, we analyzed the influence of distributed PI on short- and long-term face recognition in patients with MRI-negative epilepsy, comprising TLE, generalized epilepsy, and healthy controls. PI was induced by a list of 20 faces learned 24 hours prior to a target list of 20 faces (see Fig. 1). We tested immediate and 24-hour recognition of both lists. Thus, subjects had already looked at 100 photographs of 60 different faces before learning the target list. 2. Methods 2.1. Subjects
Fig. 1. Illustration of the study design. Proactive interference was induced by a distracter list of 20 unfamiliar faces that was learned 24 hours before a similar target list of again 20 unfamiliar faces. After the distracter list was learned, immediate and 24-hour recognition was tested. For recognition testing, 20 unknown faces were presented together with the learned faces. Immediately after the test of 24-hour recognition of the distracter list, subjects learned the target list, a new list of 20 unfamiliar faces, and they were then tested again for recognition in the same manner immediately and after 24 hours. This ‘‘distributed’’ PI paradigm differs from ‘‘massed’’ PI paradigms by stretching the presentation of the distracter information over days rather than inducing PI by presenting distracter information only immediately before the presentation of the target information. For further details, please refer to Section 2.
ence of PI on short-term memory in TLE is material specific, depending on the lateralization of the lesion. Patients with left TLE (LTLE) but not right TLE (RTLE) exhibit a detrimental effect of PI on verbal short-term memory [14–16]. Based on the dominance of the right temporal lobe in short-term face memory [17–21], we hypothesized that PI should have a detrimental effect on short-term face memory in RTLE but not in LTLE. People with TLE are a heterogeneous group, including patients with hippocampal sclerosis or other morphological alterations and lesions, as well as those with normal MRI findings. Memory disturbances in TLE may be caused
Subjects were 12 volunteers and 42 patients of the Epilepsy Center Hamburg. All participants gave informed consent to the study. All patients were undergoing a diagnostic workup with 48- to 72-hour continuous video/EEG monitoring and structural MRI investigations between April 2004 and May 2005. Dosage of medication was not lowered during video/EEG monitoring, but some of the patients were in a transition phase between different antiepileptic medication regimens. As the patients were not candidates for epilepsy surgery, reduction of medication was not necessary. Epilepsy was diagnosed when recurrent unprovoked seizures of an unquestionable epileptic nature could be verified from the patientÕs history or during video/ EEG monitoring. The diagnosis of TLE was based on reliable information about clinical seizure patterns typical of a temporal focus (aura and ictal clinical phenomena). The necessary information on seizure history and semiology was taken from interviews with the patient or witnesses of the seizures or from our own video material. During video/EEG monitoring, 11 patients exhibited left temporal interictal epileptic abnormalities (IEAs, LTLE), 9 patients exhibited right temporal IEAs (RTLE), and 10 patients had TLE with no IEAs (TLE, no IEAs) (for details on video/EEG monitoring, see the next section). Please note that patients with TLE were assigned to the RTLE or LTLE group solely on the basis of their IEAs, as most patients did not have any seizures during video/EEG monitoring (see Table 1). This method was considered valid considering the fact that other studies had found unilateral IEAs to be a good lateralizing feature of the epileptogenic region in MRI-negative TLE [25,26]. Patients without IEAs were diagnosed as having TLE only if a sufficient number of auras or seizures could be classified as typical of a temporal focus. In addition, we included 12 patients with idiopathic
222
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
Table 1 Characteristics of patients and controls
Age, mean (SD) Verbal IQ (vocabulary), mean (SD) Sex (female/male) Mean duration of epilepsy (years) Handedness (right/left) Seizures, first 24-hour delay (no/yes) Seizures, second 24-hour delay (no/yes) Psychiatric comorbidity (no/yes) IEA frequency (rare/intermittent/frequent) Sleep deprivation, first 24-hour delay Sleep deprivation, second 24-hour delay
LTLE (N = 11)
TLE, no IEA (N = 10)
RTLE (N = 9)
Generalized epilepsy (N = 12)
Controls (N = 12)
P
45 (16) 99 (8) 9/2 18 8/3 11/0 11/0 7/4 3/8/0 1 4
48 (14) 101 (9) 8/2 7 9/1 10/0 10/0 5/5 — 2 3
37 (7) 100 (8) 4/5 14 9/0 7/2 8/1 4/5 2/4/3 3 0
29 (11) 98 (7) 10/2 13 12/0 10/2 12/0 7/5 — 0 1
28 (11) 100 (11) 7/5 — 12/0 — — — — — —
<0.001a 0.90a 0.21b 0.15c 0.06b 0.14b 0.28b 0.83b 0.11b 0.08b 0.12b
LTLE, left temporal lobe epilepsy; RTLE, right temporal lobe epilepsy; IEA, interictal epileptiform abnormalities; TLE, no IEA, temporal lobe epilepsy with no IEAs detected during video-EEG monitoring; IQ, intelligence quotient. a ANOVA. b 2 v test. c Kruskal–Wallis ANOVA.
generalized epilepsy. Idiopathic generalized epilepsy was defined by typical semiology of juvenile myoclonic epilepsy, absence seizures, or primary generalized tonic–clonic seizures and generalized EEG patterns during video/EEG monitoring. Detailed structural MRI investigations were normal in all patients (for MRI details, see Section 2.3). The healthy control subjects were employees from different departments of our clinic. Among all patients, 5 had seizures and 14 were sleep-deprived for diagnostic reasons during the 24-hour delays between the two test phases (see Table 1). Seizures in TLE patients were complex partial seizures and showed ictal EEG abnormalities in accordance with the hemisphere in which IEAs had been detected. Patients with generalized epilepsy had primary generalized seizures and absence seizures. The psychiatric status of the patients was determined by patient history and clinical interviews. If necessary, questionnaires, the Beck Depression Inventory and the State–Trait–Anxiety Inventory, were used. Nineteen patients (45% of patients) received a psychiatric diagnosis in addition to epilepsy. Comorbidities included anxiety disorder (N = 6), depression (N = 10), mild psychosis (N = 1), insecure personality disorder (N = 1), and nonepileptic seizures (N = 1). 2.2. Noninvasive continuous video/EEG monitoring We employed 32- to 64-channel EEG. Electrodes were placed according to the 10/20 system, with additional electrodes according to the 10/10 system. EEG data were continuously checked for IEAs and seizure patterns by trained assistant medical technicians and evaluated by authors M.L., B.V., or S.S. Only definitive spike–slow waves were rated as IEAs. In those with TLE, IEAs were localized to electrodes F7/8, T1/2, T3/4, and T5/6. Background abnormalities or pathological slow waves were ignored for the purpose of this study. Spike–slow wave frequency was clinically rated as ‘‘none,’’ ‘‘rare,’’ ‘‘intermittent,’’ or ÔfrequentÕ
in patients with TLE (see Table 1). In patients with generalized epilepsy, generalized epileptic activity occurred mainly under hyperventilation or during sleep, and was very rare in 10 cases and occurred intermittently in 2 cases. As epileptic activity in generalized epilepsy is not directly comparable to IEAs in TLE, it was not included in Table 1. IEAs did not have to be evident during the 24-hour delays, although they were in cases with intermittent and frequent IEAs and in most of the patients with rare IEAs. However, patients were included in the left or right groups when they exhibited IEAs at any time during their stay in the EEG unit. Sleep-deprived patients had the opportunity to partly compensate for their sleep deficit before the 24-hour recognition. 2.3. Structural MRI investigations MRI scans were acquired with a 1.5-T scanner (Siemens Magnetom Symphony, Erlangen/Germany). Imaging studies included transverse T2-weighted turbo spin echo/proton density images, transverse FLAIR images, coronal T2weighted images (covering the hippocampus in a plane perpendicular to its long axis), transverse T1-weighted images, a sagittal T1-weighted MPRage three-dimensional volume scan, and transverse T1-weighted images postadministration of contrast medium. MRI studies of patients performed at other centers were reevaluated by the authors B.V., M.L., S.S., and E.G. and presented at an interdisciplinary neuroradiological case conference. In case the external MRIs did not meet the standards of the imaging protocol outlined earlier, patients were restudied at our center. 2.4. Experimental design We used 662 black-and-white face photographs (355 male), taken from a front perspective (age, 25–35). Hair was visible, but no glasses, beards, or excessive jewels.
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
Subjects received an individual set of learning and test phase items to avoid possible conversations between patients about face item characteristics. A computer program by author T.M. controlled the random draft and presentation scheme of faces and saved time of learning and test phases and the subjectsÕ answers and reaction times. During their stay in the video/EEG monitoring unit, patients learned a series of 20 unfamiliar faces presented on day 1 (distracter list); this was followed by an immediate recognition test, in which the learned faces had to be detected among 20 old and 20 new faces. After 24 hours, on day 2, recognition was tested again. After that, 20 new unfamiliar faces had to be learned (target list). Patients were tested for recognition of this second list immediately and again after 24 hours on day 3 (see Fig. 1). A 17-in. screen was employed, and photographs were 19 · 28 cm. Healthy controls were tested with a 15-in. screen, leading to a slightly smaller size of the faces presented. Before the learning phase started, subjects were verbally instructed to carefully look at the faces, as they would have to recognize them from a greater number of faces later. Twenty faces randomly picked from the face item pool were presented one at a time for 5 seconds on a computer screen. During presentation, ‘‘learning phase’’ (Lernphase) was written on the upper left corner of the screen. The first test occurred after the learning phase. An array of 40 faces in the same format were shown serially; this contained the 20 old faces randomly mixed with 20 new faces that were again randomly picked from the remaining face item pool. Subjects were instructed to decide whether they had seen the faces during the learning phase or not. Patients were informed that a second test would take place the next day. Twenty-four hours later, 40 faces (20 old faces shuffled with 20 new faces randomly taken from the face item pool) were again presented to the subjects. Subjects were instructed to decide whether they had seen the face during the learning phase the day before or not. Patients received feedback about the number of correct hits and correct rejections immediately after each test. After the 24-hour recognition test, subjects learned a new list of 20 unfamiliar faces, again randomly picked from the face item pool. Before the second learning phase started, they were informed that a new identical test would begin now, and that all faces seen so far would not be repeated anymore in the tests to come. The same round comprising learning, immediate recognition, and 24-hour delayed recognition was repeated as described earlier.
223
2.5. Statistical methods The main dependent variable was the corrected percentage of recognized faces, calculated as (hits false positives) · 100/20. We further evaluated true positive and true negative reactions and the reaction times for true positives, true negatives, false positives, and false negatives. For statistical analysis, repeated-measures analysis of covariance (RANCOVA), repeated-measures analysis of variance (RANOVA), analysis of variance (ANOVA), Kruskal–Wallis ANOVA, v2 tests, Mann–Whitney tests, and Wilcoxon tests were used. When group sizes were too small for parametric analysis (RANCOVA), we reanalyzed data with nonparametric statistics. As both methods lead to equal results and RANCOVA is more succinct in describing interaction effects, we report only these results. Significant results were further explored by specified contrasts and Scheffe´ tests. 3. Results Table 1 lists differences in clinical history and general data between the groups. Table 2 summarizes the mean face recognition performance (corrected for false positives) of the different groups. A 5 · 2 · 2 RANCOVA with the factors group (LTLE/RTLE/TLE, no IEAs/generalized epilepsy/controls), time point (immediate vs 24-hour recognition), and list (distracter list vs target list) was calculated. The covariates age, handedness, sex, sleep deprivation, and seizures during the two delays were included. As none of the covariates had a moderating influence, we report results without the inclusion of covariates. For an illustration of the results, see Fig. 2. 1. We found a significant main effect for the group factor (F = 3.9; P = 0.008). Post hoc Scheffe´ analysis revealed overall worse face recognition in patients with RTLE than in healthy controls (P = 0.01). No other groups differed from each other using post hoc analysis (P > 0.24). 2. Face recognition became worse for the target list (F = 6.9; P = 0.01). 3. Twenty-four hour recognition was worse than immediate recognition (F = 17.8; P < 0.001). 4. There was an interaction between group and time point (F = 4.2; P = 0.005). Contrast analysis revealed that healthy controls and patients with generalized epilepsy did not differ between immediate and 24-hour recognition (P > 0.32), whereas all TLE groups had worse 24-
Table 2 Mean face recognition of the different groupsa
Distractor Target a
Immediate 24-hour Immediate 24-hour
Controls
RTLE
LTLE
TLE, no IEA
Generalized epilesy
75 76 79 73
64 45 52 39
64 60 70 47
66 62 65 49
63 78 63 55
(16) (13) (15) (17)
Corrected for false positives, SD in parentheses.
(14) (13) (14) (17)
(26) (25) (16) (34)
(14) (11) (19) (23)
(19) (18) (17) (32)
224
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
Fig. 2. Face recognition of distracter and target lists. Values are means; bars denote 0.95 confidence intervals. For further details, please refer to Section 3.
hour recognition than immediate recognition (P < 0.02). In comparison, patients with generalized epilepsy showed even better 24-hour recognition than immediate recognition of the distracter face list (P < 0.001). Their 24-hour recognition of the distracter list did not differ from that of healthy controls (P = 0.74). In comparison, patients with TLE with right, left, or no IEAs were worse than healthy controls in their 24-hour recognition of the distracter list (P < 0.001, P = 0.04, and P = 0.06, respectively). 5. We found an interaction between list and time point (F = 9.3; P = 0.004). Contrast analysis showed that immediate and delayed recognition did not differ for the distracter list (P > 0.34), whereas delayed recognition was worse than immediate recognition for the target list (P < 0.001, compare Fig. 2). The latter effect was explained as follows: immediate recognition did not differ between the distracter and target lists (P = 0.92), whereas delayed recognition was worse for the target list than for the distracter list (P < 0.001; compare Fig. 2). As a number of patients had a psychiatric comorbidity, we wanted to exclude the possibility that the interaction of list and time point was caused by the psychiatric comorbidity rather than epilepsy. Thus, we tested whether the effect would still hold if we excluded patients with a psychiatric comorbidity and reanalyzed the data accordingly. The number of patients decreased to a total of 23 (for details, see Table 1). The interaction between time point and list was still visible (N = 35, P = 0.04). It is thus fairly unlikely that the psychiatric diagnosis in a number of the patients could explain the found effect. 6. The three-way interaction failed to reach significance (F = 1.8; P = 0.14). However, while patients with LTLE and generalized epilepsy did show a steeper decrease
from immediate to 24-hour recognition for the target than the distracter list (contrast analyses, P < 0.001 and P = 0.02, respectively), healthy controls, patients with TLE without IEAs, and patients with RTLE did not exhibit this effect (contrast analyses, P > 0.16). Patients with TLE without IEAs still showed a significant decrease from immediate to delayed recognition on the target list (contrast analysis, P = 0.03), but not on the distracter list (contrast analysis, P = 0.38). The RTLE group showed a decrease in immediate recognition from the first to the second list (P = 0.05), unlike all other groups (contrast analysis, P = 0.03). The RTLE group also differed from all other groups in showing a smaller decrease from immediate to delayed recognition on the target than on the distracter list (contrast analysis, P = 0.04), due to the decrease in immediate recognition of the target list. As a number of patients were sleep-deprived or had seizures during the 24-hour delays, we wanted to make sure that the interaction of list and time point was caused by epilepsy rather than these factors. We thus tested whether the effect would still hold if we excluded patients who were sleep-deprived or had seizures and reanalyzed the data accordingly. The number of patients decreased to a total of 25 (for details, see Table 1; note that one sleep-deprived patient had a seizure, and another patient had a seizure during the first and second delays). Overall, the interaction between time point and list was still apparent; i.e., longterm recognition of the target list was still worse than long-term recognition of the distracter list (N = 37, P < 0.05). The healthy controls did not show this interaction effect (contrast analysis, P = 0.39), whereas the patients with LTLE, TLE without IEAs, and generalized
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
epilepsy (N = 20) still showed the interaction effect (contrast analysis, P < 0.05). It is thus fairly unlikely that sleep deprivation or seizures in a number of patients could explain the found effect. Due to the exclusion of patients with sleep deprivation and seizures, the different patient groups were too small for further separate analysis. Although age had no influence on the results as a covariate, it might be that epilepsy patients would not differ from healthy controls were they age matched. To test this interpretation, we excluded patients aged above 55 or below 22. This led to a nonsignificant difference in age between the different groups (N = 39; F = 0.62; P = 0.65). Post hoc Scheffe´ tests were all nonsignificant as well (PÕs > 0.76). The different group sizes were still large enough for further statistical evaluation (controls: N = 7; LTLE, N = 8, RTLE: N = 9; TLE without IEAs: N = 7; generalized epilepsy: N = 8). The RANOVA was recalculated and still showed the interaction between list and time point (F = 6.4; P = 0.02). Although healthy controls did not show this interaction (N = 7; contrast analysis: P = 0.47), the patients still exhibited this effect (N = 32; contrast analysis: P = 0.02). Consistent with the initial results, this significant interaction was due mainly to the LTLE (contrast analysis: P = 0.07) and generalized epilepsy (contrast analysis: P = 0.01) groups. In summary, it is fairly unlikely that age had a moderating influence on the difference between controls and epilepsy patients in the found interaction between time point and list. To better understand the decreased delayed recognition of the target list, we reanalyzed our data by splitting the dependent variable into correctly recognized learned faces (true positives) and correctly rejected new faces (true negatives). A 5 · 2 · 2 · 2 RANOVA with the factors group (TLE left IEAs/TLE right IEAs/TLE no IEAs/generalized
225
epilepsy/controls), time point, list, and recognition type (true positives vs true negatives) was calculated. For an illustration of the results see Fig. 3. Apart from the effects already described, we found an interaction between list, time point, and recognition type (F = 6.3; P = 0.02). Contrast analysis revealed that while the number of true positives decreased from the distracter to the target list, whether immediate or 24-hour recognition was being tested (P = 0.02), the number of true negatives differed between the distracter and target lists, depending on whether it was immediate or 24-hour recognition. Although the number of true negatives during immediate recognition was higher for the target than the distracter list (contrast analysis, P = 0.02), it was lower for the target than the distracter list during 24-hour recognition (contrast analysis, P < 0.001) (see Fig. 2). While the number of true negatives from the distracter list did not differ between immediate and 24-hour recognition (P = 0.31), fewer true negatives were visible during delayed than immediate recognition (P < 0.001). No three-way interaction was visible (P = 0.73). In summary, the detrimental effect of PI on 24-hour face recognition is due mainly to a combined decrease of true positive and true negative reactions during 24-hour recognition. We also analyzed reaction times for true positive, true negative, false positive, and false negative responses. A 5 · 2 · 2 · 4 RANOVA with the factors group (TLE left IEAs/TLE right IEAs/TLE no IEAs/generalized epilepsy/ controls), time point, list, and reaction time (true positive, true negative, false positive, and false negative responses) was calculated. Mean reaction times increased during delayed recognition trials (immediate: 3.2 seconds, SE = 0.18; 24-hour: 3.6 seconds, SE = 0.21; F = 5.0; P = 0.03). False responses took longer than correct
Fig. 3. True positive and true negative reactions during immediate and delayed recognition of the distracter and target lists. Values are means; bars denote 0.95 confidence intervals. For further details, please refer to the Section 3.
226
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
responses (mean: false positives: 4.3 seconds, SE = 0.37; false negatives: 3.7 seconds, SE = 0.25; true positives: 2.6 seconds, SE = 0.10; true negatives: 3.0 seconds, SE = 0.13; F = 31.5; P < 0.001). False negative and false positive responses did not differ (F = 2.5; P = 0.12), but true positive responses were faster than true negative responses (F = 13.0; P < 0.001). No other effects were detected (P P 0.12). To better understand the face recognition deficit in RTLE, we analyzed the correlations between 24-hour recognition of the distracter list and that of the target list. The correlations were as follows: healthy controls, r = 0.82 (P = 0.01); RTLE, r = 0.25 (P = 0.51); LTLE, r = 0.65 (P = 0.03); TLE, no IEAs, r = 0.34 (P = 0.34); generalized epilepsy, r = 0.27 (P = 0.47). The correlations between RTLE and healthy controls differed (P = 0.02, two-tailed). No other groups differed from each other in this respect. 4. Discussion The present study analyzed the influence of distributed PI on short- and long-term face memory in patients with MRI-negative TLE, patients with generalized epilepsy, and healthy controls. The main results of the study are that (1) PI led to decrease in 24-hour delayed recognition, but not to a decline in immediate recognition; (2) this effect was significant in patients with LTLE, patients with generalized epilepsy, and, to a lesser degree, in patients with TLE without IEAs (in comparison, it was not significant in healthy controls and patients with RTLE); and (3) in opposition to the other groups, PI led to decreased short-term memory in patients with RTLE. To our knowledge, this study demonstrates for the first time a heightened effect of PI on long-term recognition in epilepsy patients. Our results suggest that although a distributed PI paradigm does not decrease 24-hour recognition in healthy subjects [4], distributed PI may be a source of long-term memory disturbance in epilepsy patients. The results complement studies that have found an influence of massed PI on 24-hour long-term memory in healthy individuals [3,4]. The mechanisms underlying PI are not known. However, evidence has been found for a role of the prefrontal cortex and frontopolar cortex in resolving the effect of PI in short-term recognition [27–29]. In our study, LTLE and generalized epilepsy were related to a stronger effect of PI on memory than TLE without IEAs or healthy controls, suggesting that prefrontal lobe function is negatively influenced in LTLE or generalized epilepsy. This conjecture is supported by studies that found frontal and prefrontal dysfunction in LTLE and generalized epilepsy [e.g., [30–33]]. Our result that PI affects 24-hour recognition but not immediate recognition is in agreement with studies in healthy subjects, in whom the effect of PI on memory was observed to increase with the passage of time between learning and recall [e.g., [34]].
The present study also shows for the first time that PI affects immediate face recognition in RTLE but not LTLE. This finding complements studies that reported a detrimental effect of PI on verbal short-term memory in LTLE [14– 16]. Interestingly, PI did not lead to a further decrease in 24-hour delayed recognition in patients with RTLE, as was visible in LTLE, generalized epilepsy, and TLE without IEAs. Perhaps the distracter list causes less disturbance in the recognition of faces from the target list 24 hours later, because patients with RTLE already exhibited a deficit in recognition of the distracter list. The significantly lower and negative correlation between 24-hour recognition of the distracter list and that of the target list in RTLE compared with the high positive correlation in healthy controls supports this interpretation. However, the present study could not determine whether PI has an effect on long-term face recognition in RTLE. This can only be investigated if performance in the recognition of the distracter list is matched in all groups, as was done in studies of healthy controls [3]. Rather, our study suggests that it may be more difficult to further induce a detrimental effect of PI on the already decreased long-term face memory in RTLE. That patients with RTLE had an overall worse face memory compared with healthy controls confirms the dominant role of the right temporal lobe in face memory [17–21]. In the present study, PI was elicited by learning a list of 20 faces 24 hours before learning a second list of 20 faces, in accordance with a distributed PI paradigm (compare [4]). Immediately before learning the target list, subjects were again presented with 40 faces for testing 24-hour recognition of the distracter list. It is hence not possible to distinguish whether the PI effect on long-term memory evident in this study was caused by learning the distracter list, or by its presentation during the 24-hour recognition prior to learning of the target list, or by a combination of both. However, the aim of the present study was to establish whether epilepsy heightens the effect of PI on long-term memory. Future studies can determine under what conditions this is the case. The influence of PI on memory as seen in the patients studied might be a nonspecific effect of antiepileptic medication, which was not controlled for in this study. A recent study found accelerated forgetting in patients with high serum levels of antiepileptic medications [35]. Another study found that differences between LTLE and RTLE with respect to the effect of proactive and retroactive interference on verbal short-term memory were visible only under full medication, but not under moderately reduced anticonvulsant medication [14]. However, as most epilepsy patients usually depend on anticonvulsant medication, the present study at least proves that distributed PI has a specific impact on short-term and long-term face memory in epilepsy patients under their regular medication regimen. Our study was restricted to patients with MRI-negative epilepsy. Although this restriction raised the homogeneity of the studied sample and made the comparison of patients with TLE, patients with generalized epilepsy, and healthy
T. Bengner et al. / Epilepsy & Behavior 8 (2006) 220–227
controls easier, it is not known whether lesional epilepsy patients will exhibit the same susceptibility to PI. A majority of patients with TLE show hippocampal sclerosis or other lesions. Thus it is up to future studies to prove whether PI decreases short- and long-term face recognition in MRI-positive, lesional TLE. Based on our observation that PI impairs memory in epilepsy patients, the results of this study might be clinically relevant. For example, face memory tests eliciting PI in combination with longer testing delays between learning and recognition could increase the sensitivity of detecting memory disturbances in epilepsy, e.g., when patients complain about memory deficits in everyday life but perform normally on standard memory tests. References [1] Keppel G, Underwood BL. Proactive inhibition in short-term retention of single items. J Verb Learn Verb Behav 1962;1:153–61. [2] Shimamura AP, Jurica PJ, Mangels JA, Gershberg FB. Susceptibility to memory interference effects following frontal lobe damage: findings from tests of paired-associate learning. J Cogn Neurosci 1995;7:144–52. [3] Underwood BL. Interference and forgetting. Psychol Rev 1957;64:49–60. [4] Underwood BJ, Ekstrand BR. An analysis of some shortcomings in the interference theory of forgetting. Psychol Rev 1966;73:540–9. [5] Wixted JT. The psychology and neuroscience of forgetting. Annu Rev Psychol 2004;55:235–69. [6] Yarmey AD. Proactive interference in short-term retention of human faces. Can J Psychol 1974;28:333–8. [7] Flicker C, Ferris SH, Crook T, Bartus RT. Age differences in the vulnerability of facial recognition memory to proactive interference. Exp Aging Res 1989;15:189–94. [8] Warrington EK, Weiskrantz L. The effect of prior learning on subsequent retention in amnesic patients. Neuropsychologia 1974;12:419–28. [9] Warrington EK, Weiskrantz L. Further analysis of the prior learning effect in amnesic patients. Neuropsychologia 1978;16:169–77. [10] Moscovitch M, Winocur G. Contextual cues and release from proactive inhibition in young and old people. Can J Psychol 1983;37:331–44. [11] Milner B. Psychological aspects of focal epilepsy and its neurosurgical management. In: Purpura D, Penry J, Walter R, editors. Advances in neurology: vol. 8. Neurosurgical management of the epilepsies. New York: Raven Press; 1975. p. 299–321. [12] Ladavas E, Umilta C, Provinciali L. Hemisphere-dependent cognitive performances in epileptic patients. Epilepsia 1979;20:493–502. [13] Morris RG, Polkey CE, Cox T. Recognition memory for words and faces following unilateral temporal lobectomy. Br J Clin Psychol 1995;34:571–6. [14] Durwen HF, Elger CE. Verbal learning differences in epileptic patients with left and right temporal lobe foci: a pharmacologically induced phenomenon? Acta Neurol Scand 1993;87:1–8. [15] Giovagnoli AR, Avanzini G. Forgetting rate and interference effects on a verbal memory distractor task in patients with temporal lobe epilepsy. J Clin Exp Neuropsychol 1996;18:259–64.
227
[16] Winocur G, Moscovitch M, Bruni J. Heightened interference on implicit, but not explicit, tests of negative transfer: evidence from patients with unilateral temporal lobe lesions and normal old people. Brain Cogn 1996;30:44–58. [17] Milner B. Visual recognition and recall after right temporal-lobe excision in man. Neuropsychologia 1968;6:191–209. [18] Barr WB. Examining the right temporal lobeÕs role in nonverbal memory. Brain Cogn 1997;35:26–41. [19] Kelley WM, Miezin FM, McDermott KB, et al. Hemispheric specialization in human dorsal frontal cortex and medial temporal lobe for verbal and nonverbal memory encoding. Neuron 1998;20:927–36. [20] Kelley WM, Ojemann JG, Wetzel RD, et al. Wada testing reveals frontal lateralization for the memorization of words and faces. J Cogn Neurosci 2002;14:116–25. [21] Coleshill SG, Binnie CD, Morris RG, et al. Material-specific recognition memory deficits elicited by unilateral hippocampal electrical stimulation. J Neurosci 2004;24:1612–6. [22] Giovagnoli AR, Avanzini G. Learning and memory impairment in patients with temporal lobe epilepsy: relation to the presence, type, and location of brain lesion. Epilepsia 1999;40:904–11. [23] Baxendale SA. The role of the hippocampus in recognition memory. Neuropsychologia 1997;35:591–8. [24] Glogau S, Ellgring H, Elger CE, Helmstaedter C. Face and facial expression memory in temporal lobe epilepsy patients: preliminary results. Epilepsy Behav 2004;5:106–12. [25] Holmes MD, Born DE, Kutsy RL, Wilensky AJ, Ojemann GA, Ojemann LM. Outcome after surgery in patients with refractory temporal lobe epilepsy and normal MRI. Seizure 2000;9:407–11. [26] Sylaja PN, Radhakrishnan K, Kesavadas C, Sarma PS. Seizure outcome after anterior temporal lobectomy and its predictors in patients with apparent temporal lobe epilepsy and normal MRI. Epilepsia 2004;45:803–8. [27] Jonides J, Marshuetz C, Smith EE, Reuter-Lorenz PA, Koeppe RA, Hartley A. Age differences in behavior and PET activation reveal differences in interference resolution in verbal working memory. J Cogn Neurosci 2000;12:188–96. [28] Badre D, Wagner AD. Frontal lobe mechanisms that resolve proactive interference. Cereb Cortex 2005;15:2003–12. [29] King JA, Hartley T, Spiers HJ, Maguire EA, Burgess N. Anterior prefrontal involvement in episodic retrieval reflects contextual interference. Neuroimage 2005;28:256–67. [30] Jokeit H, Seitz RJ, Markowitsch HJ, Neumann N, Witte OW, Ebner A. Prefrontal asymmetric interictal glucose hypometabolism and cognitive impairment in patients with temporal lobe epilepsy. Brain 1997;120:2283–94. [31] Swartz BE, Simpkins F, Halgren E, et al. Visual working memory in primary generalized epilepsy: an 18FDG-PET study. Neurology 1996;47:1203–12. [32] Holmes MD, Brown M, Tucker DM. Are ‘‘generalized’’ seizures truly generalized? Evidence of localized mesial frontal and frontopolar discharges in absence. Epilepsia 2004;45:1568–79. [33] Zifkin B, Andermann E, Andermann F. Mechanisms, genetics, and pathogenesis of juvenile myoclonic epilepsy. Curr Opin Neurol 2005;18:147–53. [34] Briggs GE. Acquisition, extinction, and recovery functions in retroactive inhibition. J Exp Psychol 1954;47:285–93. [35] Jokeit H, Kramer G, Ebner A. Do antiepileptic drugs accelerate forgetting? Epilepsy Behav 2005;6:430–2.