Cognitive Brain Research 12 (2001) 89–99 www.elsevier.com / locate / bres

Research report

Frontal-parietal activation differences observed before the execution of remembered saccades: an event-related potentials study 1

Ioannis Evdokimidis*, Nikolaos Smyrnis, Theodoros S. Constantinidis, Pavlos Gourtzelidis , Costantinos Papageorgiou Cognition Action Group, Neurology Department, National University of Athens Medical School, Aeginitio Hospital, Vas. Sofias 72 -74, Athens GR-11528, Greece Accepted 20 February 2001

Abstract Healthy subjects performed saccadic eye movements in one memory (MEM) and two delay tasks (delay, DEL and modified delay, M-DEL) while we recorded scalp event-related potentials (ERPs) from 25 electrode sites. In the MEM task the subjects were instructed to retain in memory the location of a visual target for a delay of 1–6 s and then perform a remembered saccade at the go signal. In the DEL task the target remained on until movement completion and in the M-DEL task the target, that was visible during the delay period, disappeared synchronously with the go signal. A reduction in response latency and an increase in the percentage of dysmetric movements were observed for the MEM task compared to the two delay tasks. An increased ERP activity at the central-frontal electrode sites compared to the parietal sites was significant only for the MEM task early on during the delay period (500–1000 ms). During the period preceding the onset of the saccade, a parietal increase of activity was observed for all tasks. Furthermore the activity was smaller for the frontal compared to the parietal areas only for the memory task thus indicating a near reversal of the previous pattern of activity observed during the early delay period. This specific activation pattern of frontal and parietal areas, observed for the MEM task only, requires further investigation focusing on the temporal pattern of activation of large brain areas involved in working memory processing.  2001 Elsevier Science B.V. All rights reserved. Theme: Motor systems and sensorimotor integration Topic: Oculomotor systems Keywords: Event-related potential; Working memory; Saccade; Eye-movement; Frontal cortex; Parietal cortex

1. Introduction The pattern of cortical activation before the generation of a saccade seems to depend on the type of the intended eye movement. Simple, visually guided reflex saccades involve the activation of the parietal cortex [1,16]. This activation is probably reflected in studies of scalp recorded event-related potentials (ERPs) in humans. Specifically, self-paced horizontal saccades are preceded by a midline activation starting at about 1 s prior to the saccade onset while visually triggered saccades are preceded by an *Corresponding author. Tel.: 130-1-7289-115; fax: 130-1-7216-474. E-mail address: [email protected] (I. Evdokimidis). 1 Present address: Neurophysiology Department, 401 General Military Hospital, Athens, Greece.

activity of similar topography but of shorter duration [20]. Although the origin of these potentials is not entirely clear, it is well established that the EEG changes depend on the type of saccade [6]. For example, the ERPs preceding the execution of anti-saccades (eye movements in the opposite direction from a visual cue) reveal a frontal activation in addition to the parietal activity [7]. This finding is in accordance with the fact that frontal patients have difficulty in inhibiting pro-target saccades in order to perform anti-saccades [18]. Thus, the use of scalp-recorded ERPs in humans is a method with high temporal resolution for the study of cognitive operations such as inhibition of a reflexive response in order to execute a volitional eye movement. The remembered saccade, i.e., saccade towards a remembered visual target in space also involves a set of

0926-6410 / 01 / $ – see front matter  2001 Elsevier Science B.V. All rights reserved. PII: S0926-6410( 01 )00037-4

90

I. Evdokimidis et al. / Cognitive Brain Research 12 (2001) 89 – 99

cognitive operations before the final saccade execution. Like the anti-saccades, remembered saccades require the inhibition of the reflexive pro-target movement. In addition, the performance of a remembered saccade requires the maintenance of the target coordinates in working memory for a period before the saccade is executed. Studies in animals have already shown that neurons in prefrontal [10–12] and posterior parietal cortices [2– 4,15,30], fire in relation to both the target as well as the direction of the intended saccade, indicating that these neurons participate in working memory functions. Lesions in humans involving the anterior areas and especially the dorsolateral prefrontal cortex result in the deterioration of the performance of remembered saccades as shown by an increase in dysmetria and latency of these eye movements [26]. Recently, functional brain imaging studies showed that several cortical and subcortical areas participate in spatial working memory tasks and, in spite of the complexity of activation patterns, the participation of the dorsolateral prefrontal and posterior parietal cortices seems to be a rather constant finding [19,24,31]. The bi-directional anatomical connections between the posterior parietal and dorsolateral prefrontal cortices support the hypothesis of a functional connectivity between these areas [17]. Whereas the study of remembered saccades in humans using functional brain imaging, can reveal the anatomical characteristics of this network, the temporal dynamics of the network can be more appropriately investigated with the use of ERPs that have a better time resolution. In a recent study ERPs were recorded before the execution of remembered saccades [27]. This study focused only on the EEG changes occurring during the memorization period and revealed a frontal-central activation that was more pronounced with the memory saccades. In this study we investigated the cortical potentials preceding delay and remembered saccades analyzing the EEG changes occurring during the delay as well as during the presaccadic period. More specifically, we compared the cortical activation among three tasks in three time windows. The first task (MEM) required the retention of the target location in working memory for a delay period of 1–6 s. In the second task (DEL) the target remained visible during the delay, the pre-saccadic period and until movement completion while in the third task (M-DEL) the target disappeared during the pre-saccadic period. We compared ERP activity for the three tasks during the early delay, late delay and pre-saccadic periods.

2. Materials and methods Twelve healthy subjects participated in the experiment. Subjects were recruited from the academic environment of the Aeginition University Hospital. All subjects gave their informed consent. The Scientific and Ethics committee of the Aeginition University Hospital approved the ex-

perimental protocol. Subjects used corrected vision where appropriate.

2.1. Stimulation procedure Subjects sat at a distance of 100 cm in front of a gray light monitor screen where the visual stimuli were presented. The experimental paradigm consisted of three types of trials presented in a random order (Fig. 1). For all three types, the trial started with the appearance of a central target (white cross 0.530.58) in the middle of the screen. Subjects were instructed to fixate this central target. After a randomly variable period of 1–2 s, a peripheral target (same configuration) appeared, either to the left or to the right with an eccentricity of 6 or 128 (each one of the four target locations was randomly chosen for each trial), indicating the end point of the saccade. After another randomly variable period of 1–6 s, the central target disappeared (GO signal) indicating to the subjects to make a saccade to the peripheral target location. The central target reappeared after a variable period of 2–4 s, indicating the start of a new trial. Each subject had a practice session of a few trials in order to master the task.

2.1.1. The delay task, DEL The peripheral target remained on during the entire trial. Specifically, the target was visible during the delay period of 1–6 s preceding the GO signal, during the subsequent 1-s period during which the saccade was initiated, and during another variable 1–3-s period before the initiation of a new trial. 2.1.2. The modified delay task, M-DEL In contrast to the delay task, the peripheral target did not remain on during the entire trial. Instead, the peripheral target disappeared simultaneously with the central target (GO signal), requiring the subject to remember the location of the target at the time of saccade initiation. One second after the GO signal, a correction target appeared and the subject could perform a corrective saccade if needed. 2.1.3. The memory task, MEM In contrast to the two delay tasks, the peripheral target did not remain on during the entire period preceding the GO signal. Instead, it flashed for 200 ms and then the ‘memory’ period started, lasting from 1 to 6 s, requiring the subject to remember the peripheral target location for an extended period of time prior to saccade initiation. At the end of the memory period, the central target disappeared (GO signal) indicating to the subjects to make a remembered saccade to the peripheral target location. As in the modified delay task, after 1 s a correction target appeared in the location of the previously presented peripheral target and the subject could perform a corrective saccade if necessary.

I. Evdokimidis et al. / Cognitive Brain Research 12 (2001) 89 – 99

91

Fig. 1. Experimental setup. The experimental paradigm is presented in this figure. The timing diagrams, in order of presentation in the figure, are as follows: (a) central target, common for all tasks, which indicates the start of a new trial when turned on and the signal to move when turned off; (b) the peripheral target for the three tasks: DEL task — the peripheral target remains on throughout the trial, M-DEL task — the peripheral target is extinguished simultaneously with the central target (GO signal) and reappears after 1 s (correction target), MEM task — the peripheral target is presented as a 200-ms flash at the beginning of a trial and reappears 1 s after the GO signal (correction target); and (c) a hypothetical EOG record, illustrating the onset of a saccade. The three periods of analysis (early delay period, late delay period, pre-saccadic period) are shown in gray across all of the above.

2.2. Recordings The EOG was recorded using the infrared method (Iris, Skalar  ). The signal from the right eye was used to measure the horizontal eye movement component and the signal from the left eye was used to measure the vertical component in order to detect various types of blinks. The EEG was recorded using a conventional technique [7] with surface electrodes over 25 leads (10 / 20 I.S., F 3 , F 1 , Fz, F 2 , F 4 , CF 3 , CF 1 , CFz, CF 2 , CF 4 , C 3 , C 1 , Cz, C 2 , C 4 , CP3 , CP1 , CPz, CP2 , CP4 , P3 , P1 , Pz, P2 , P4 ). The EEG and EOG signals were digitized at 200 Hz and were stored together with the stimulus information on the hard drive for subsequent off-line analysis.

2.3. Data analysis 2.3.1. Behavior An investigator inspected the EOG signal for each trial and excluded trials with artifacts (blinks, eye movements larger than 18 during the delay period). He then used an

interactive program to measure the first change in the EOG record indicating the onset of a saccade. Response latency was determined from the EOG measurement as the first eye movement after the GO signal. The percentage of dysmetria was calculated from the saccade amplitude (measured in degrees) according to the following formula: [(expected amplitude2performed amplitude) / expected amplitude]3100 [26]. When the percentage of dysmetria was below 10% we considered the movement as accurate. It has been reported that such a small percentage of dysmetria (usually hypometria) is frequently encountered in normal individuals [21]. All other movements were considered as dysmetric. For the statistical analysis of the latencies, we used an ANOVA with latency as the dependent variable and task type, delay and dysmetria (binary variable) as the independent variables. The delays in this analysis were grouped in five categories of 1 s delay duration each (1–2 s, 2–3 s, etc.). For post hoc analyses we used the Tukey Honest Significance test with the significance level set at 0.05. For the statistical analysis of the dysmetria, we used a

92

I. Evdokimidis et al. / Cognitive Brain Research 12 (2001) 89 – 99

logistic regression model with three independent variables, one describing the contrast of the MEM task with the two delay tasks (binary variable), one describing the contrast between the two delay tasks (binary variable) and finally the delay (parametric variable). The dysmetria was the dependent variable (binary variable).

2.3.2. ERPs Each subject performed a set of approximately 100–120 trials. From the original data set we excluded all trials with artifacts that were detected with visual inspection of the EEG and EOG data. Most of the times the artifacts were due to blinks or saccadic eye movements during either the delay period, or during the last 500 ms before the cue presentation or during the cue period. Thus we retained a set of 811 trials for all subjects that were artifact free. For the statistical analysis of the ERPs we excluded all trials with response latencies above 600 ms (n540) and from the remaining trials we excluded all those with dysmetria. Thus from the original set of 811 trials from all subjects pooled together we used 573 trials (197 trials for the MEM task, 177 trials for the DEL task and 199 trials for the M-DEL task). We divided the time series data for each trial in three time periods as follows:

(A) The first period, or ‘early delay period’, consisted of 700–1200 ms after the peripheral target appearance, corresponding to 500–1000 ms after the disappearance of the peripheral target in the MEM task. We subtracted the average activity during the 500 ms preceding peripheral target appearance from the recorded signal, taking this to represent baseline activity. (B) The second period, or ‘late delay period’, consisted of the 500 ms prior to the GO signal for all three tasks. The baseline activity that was subtracted from the signal was again the average activity during the 500 ms preceding the peripheral target appearance as for the early delay period. (C) The third period, or ‘pre-saccadic period’, consisted of the 200 ms preceding saccade onset for all three tasks. We subtracted the average activity during the 50 ms preceding the GO signal. The selection of this baseline activity helped to magnify the specific signal differences between the tasks that follow the GO signal because any differences in signal preceding the GO signal have already been investigated in the late delay period.

successively at 50-ms intervals using the average of five signal values centered at the particular time (for example at 50 ms). Thus, for the early delay period, we retained 10 signal values for each trial, from 700 to 1150 ms after the peripheral target appearance at 50-ms intervals). For the late delay period, we also retained 10 signal values for each trial, from 500 to 50 ms before the GO signal at 50-ms intervals. Finally, for the pre-saccadic period we retained four values for each trial, from 200 to 50 ms before the saccade onset at 50-ms intervals. We performed three repeated measures ANOVAs, one for each time period, with time as the repeatedly measured dependent variable and task, delay duration (only in the analysis of the late delay period) and electrode group as the independent variables (fixed factors). Due to the fact that data from different subjects were pooled for this analysis and the contribution of each subject to the total activity might be different depending on external factors such as scalp conductance we used a Mixed model ANOVA and introduced the subjects as a random effects factor. This way we could dissociate in our analysis the variance due to the fixed factors that were experimentally manipulated (type of task channel group and delay) from the variance due to the random effects introduced by the particular selection of this group of subjects. The underlying assumption is that this group is a random selection from a population of healthy subjects. In our data set for the ERP analysis the subject with the smallest number of trials contributed 34 trials and the subject with the largest number of trials contributed 82 trials. One subject that had only 21 valid trials was excluded from the ERP ANOVA analysis. Delay durations were grouped into two categories, short delay from 1 to 3 s and long delay of 4–6 s. Five electrode groups were allocated as follows: frontal (F 3 , F 1 , Fz, F 2 , F 4 ), central-frontal (CF 3 , CF 1 , CFz, CF 2 , CF 4 ), central (C 3 , C 1 , Cz, C 2 , C 4 ), central-parietal (CP3 , CP1 , CPz, CP2 , CP4 ) and parietal (P3 , P1 , Pz, P2 , P4 ). For post-hoc analyses we used the Tukey honest significance test and the significance level was set to 0.01. For visualization of ERP data in Figs. 3 and 5 the single data waveforms were averaged for all subjects for each task condition (grand average waveforms) after subtraction of the baseline activity. No smoothing was applied to these waveforms.

3. Results

3.1. Behavior It should be noted that for the 1-s delay interval the early memory and the late memory periods coincided. Since the three periods were analyzed separately, a small subset of the data was included in both early and late delay period analyses. The average value of the EEG signal for each trial, for each period and for each electrode site was derived

There was an increase in the percentage of dysmetric movements in the MEM task compared to the DEL and M-DEL tasks (Table 1). In fact, the regression factor contrasting the MEM task versus the two delay tasks was a reasonably good predictor of dysmetria (Wald x 2 54.02, P,0.045) while the factor contrasting the two delay tasks

I. Evdokimidis et al. / Cognitive Brain Research 12 (2001) 89 – 99 Table 1 Dysmetria for the three tasks

MEM (%) DEL (%) MDEL (%) Total (%)

93

3.2. Event-related potentials ( ERPs)

1–2 s

2–3 s

3–4 s

4–5 s

5–6 s

Total

23.1 14.9 30.0 21.2

30.1 16.6 10.5 19.6

25.5 40.5 27.0 30.4

29.4 26.9 23.4 26.2

37.3 26.0 27.9 31.1

29.6 24.3 22.6 25.7

The percentages of dysmetric movements are presented in this table by task (rows) and delay period (columns). The delays from 1 to 6 s were grouped in five delay periods for presentation purposes. The dysmetria for the memory task (MEM) was larger than that of the delay tasks (DEL, M-DEL) (see row totals) and the dysmetria increased with delay (column totals).

was not significant. There was also an increase in dysmetria at longer delays (Table 1) and the delay was a strong predictor of dysmetria (Wald x 2 57.69, P,0.005). There was a significant effect of task on saccadic response latency (F2 58.72, P,0.001). Post hoc analysis confirmed a significantly shorter latency for the MEM task compared to the latency for the DEL task and the M-DEL task, while the difference between the DEL and M-DEL tasks was not significant (Fig. 2A).There was also a significant effect of delay on saccadic response latency (F4 514.71, P,0.001). Post hoc analysis of the delay effect confirmed a significant decrease in latency with increasing delay. The latencies to dysmetric movements were not significantly different from those for correct movements (Fig. 2C). Finally no significant interactions between different factors were observed. In summary, the percentage of dysmetric movements was larger and the response latency was shorter for the MEM task compared to the two delay tasks. The dysmetria and the latency of response were not different between the two delay tasks.

3.2.1. Early delay period Fig. 3A depicts the average ERPs for all subjects for the early delay period (gray crosshatch). Immediately preceding the early delay period, the extinction of the peripheral target for the MEM task resulted in a negative potential with a peak at about 150 ms (the response to the disappearance of the visual target) followed by a P300 -like positive deflection. During the early delay period there was a significant difference in ERPs among electrode groups (F4 512.23, P,0.0001), (Fig. 4A). The post hoc analysis showed that the ERP for the central-frontal group was significantly larger from that of all other electrode groups. In addition, there was a significant interaction between task and electrode group (F8 52.55, P,0.05). In a post-hoc analysis we investigated this interaction by comparing (a) the activity for different tasks for each electrode group and (b) the activity for each task across electrode groups. The first set of comparisons between tasks for different electrode groups did not show any significant differences. The second group of comparisons across electrode groups showed for the MEM task: (a) the frontal group ERP was significantly larger than that of the parietal group; (b) the central-frontal group ERP was significantly larger than that of the central-parietal and parietal groups; and, finally, (c) the central group ERP was significantly larger than that of the parietal group ERP. For the DEL task no difference was significant, while for the M-DEL task only the centralfrontal electrode group ERP was significantly larger than that of the parietal electrode group (Figs. 3B and 4B). Thus an increase in activity over central and frontal areas was observed for the MEM task when activity in these areas was compared to the activity over the parietal area.

Fig. 2. Response latencies for the three tasks. (A) The mean saccadic response latencies for the three tasks pooled across delays. (B) The mean latencies for the five delay intervals pooled across tasks (see Section 2). (C) The mean latencies for the correct and dysmetric movements pooled across delays and tasks. The error bars for each histogram depict standard deviations.

94

I. Evdokimidis et al. / Cognitive Brain Research 12 (2001) 89 – 99

Fig. 3. ERPs for the early delay period. (A) The average ERPs for the memory task (thick solid line), the delay task (thin solid line) and the modified delay task (dotted line) for 25 electrode sites on the scalp (the C z , Pz and F z leads are marked) for all subjects (grand average). According to convention, negativity is depicted as a positive deflection (uV5microvolts). The ERPs for the first second after the extinction of the peripheral target are depicted in this figure. The horizontal lines at each plot represent the average baseline activity. The last 500 ms of this period was the early delay period, marked with a gray strip. (B) The subtracted ERPs for the difference between CF 1 and P1 and the difference between CF z and Pz electrodes for each task using the same types of lines in the same time interval.There is a large increase in activity for the MEM task compared to the two delay tasks that is more prominent during the early delay period (gray strip).

3.2.2. Late delay period The difference between frontal and parietal electrode group ERPs remained during the late delay period (significant electrode group effect, F4 55.68, P,0.01), although the overall ERP activity was reduced (Fig. 4A). More specifically the post hoc analysis showed that the activity for the parietal group was significantly smaller from that of the central-frontal group and the central group. However the task by electrode group interaction was not significant in contrast to the early delay period. The duration of the delay did not affect activity nor were there any significant interactions between delay and task. 3.2.3. Pre-saccadic period Fig. 5A depicts the average ERPs recorded for the three tasks before the onset of saccadic eye movement (pre-

saccadic period). A significant main effect of electrode group was observed (F4 52.77, P,0.05). More specifically the post hoc analysis showed that the frontal group activity was significantly smaller than that of all other groups (see Fig. 4C). Furthermore, there was a significant interaction between task and electrode group (F8 52.13, P,0.05). We further investigated this interaction in a post-hoc analysis by comparing (a) the activity across different tasks within each electrode group and (b) the activity within each task across electrode groups. The comparison within electrode groups across tasks revealed that the activity for the DEL task was significantly larger than the activity for MEM task for all electrode groups except the parietal where activity for all tasks was equivalent. Furthermore the activity for the M-DEL task was significantly smaller than that for the DEL task for all groups except the central-

I. Evdokimidis et al. / Cognitive Brain Research 12 (2001) 89 – 99

95

4. Discussion In this study we recorded ERPs while normal healthy volunteers performed saccadic eye movements in one memory (MEM task) and two delay (DEL, M-DEL tasks) conditions. The main findings from this study could be summarized as follows:

Fig. 4. ERP average task differences. (A) The mean activity (negativity is plotted as positive y values) for each electrode group for the early delay period (solid line) and the late delay period (dotted line). The plot depicts the activity averaged for all tasks. A clear decrease in overall activation is observed for the late delay period although the activity is still larger for the central-frontal electrode group compared to the parietal group. (B) The activity for each task separately during the early delay period (thick line, MEM; thin line, DEL; and dotted line, M-DEL). A clear difference in negativity is observed for the frontal-central electrode group versus the parietal group and this difference is larger for the MEM task. (C,D) The mean activity for each electrode group for all tasks and each task separately, respectively, during the pre-saccadic period. A pattern of activity difference between frontal and parietal groups (frontal smaller than parietal) is particularly evident for the MEM task.

(I) The performance of remembered saccades was preceded by shorter latencies and resulted in a larger percentage of dysmetric movements. The disappearance of the peripheral target after the go signal (MDEL task) did not account for this behavioral difference between remembered and delay saccades because in that case the latency and dysmetria was not different from that of visually guided delay saccades (DEL task). (II) During the early delay period a pattern of larger activity over the frontal areas compared to the parietal areas was observed that was specific for the memory task. During the late delay the same pattern of frontal versus parietal increase in activity remained but it was not specific for the memory task. (III) The ERP activity preceding the onset of the saccadic eye movement was larger over the parietal areas for all three tasks. A spread of this activity to central and frontal areas was specific to the DEL task. More interestingly the comparison of activity for each task across electrode sites revealed a significant decrease in activity for frontal versus parietal areas only for the memory task. This pattern of frontal versus parietal difference of activity was a near reversal of the pattern observed for the same task during the early delay period.

4.1. Early delay activity: evidence for a memory-related activation pattern parietal and parietal. Finally the activity for the M-DEL and MEM tasks was not significantly different for all electrode groups. The second comparison within each task and across electrode groups showed for the MEM task that the frontal electrode group activity was significantly smaller from that of the central-parietal and parietal group (Figs. 4D and 5B). The activity for the other two tasks did not differ significantly across electrode groups. In summary, during the presaccadic period a parietal activity was observed for all tasks. This activity spread over the central and frontal areas for the DEL task only. Furthermore a pattern of difference in activity between frontal and parietal areas was observed only for the MEM task. Interestingly this pattern was qualitatively the reverse from the pattern observed for the same task during the early memory period namely an increase of activity over the parietal areas compared to the activity over the frontal and central areas.

A large body of literature has already shown that when a subject performs a delayed response then a frontal activation appears. The cognitive operation performed involves both the inhibition of the reflex response as well as the maintenance in working memory of the target location and / or the motor program for the intended delayed action. Neuronal activity in the prefrontal cortex (PFC) of the monkey is related both the target location and the intended eye movement response in the contralateral space [10,13]. The performance on visuospatial delay tasks is also severely impaired with lesions of certain prefrontal structures such as the dorsolateral PFC [26]. A frontal activation involvement in spatial working memory tasks was also revealed in neuroimaging studies [19]. When remembered saccades were investigated using functional brain imaging several sites of activation were observed, such as, the dorsolateral PFC, the frontal eye fields (FEF), the supple-

96

I. Evdokimidis et al. / Cognitive Brain Research 12 (2001) 89 – 99

Fig. 5. ERPs for the pre-saccadic period. (A) The average ERPs recorded form 25 electrode sites (the Fz, Cz and Pz leads are marked) for all subjects (grand averages) before the execution of the saccade. Again negativity is depicted as a positive deflection (uV5microvolts). The horizontal lines at each plot represent the average baseline activity. The thin solid line in each diagram represents the ERP for the delay task (DEL), the thin dotted line represents the ERP for the modified delay task (M-DEL) and the thick solid line represents the ERP for the memory task (MEM). It can be seen that a negative potential appears just before the saccadic eye movement. This potential is larger over the central and parietal electrode leads and is larger for the DEL task, smaller for the M-DEL task and even smaller for the MEM task (except for the parietal leads where the difference between tasks is not significant). (B) The subtracted ERPs for the difference between F 1 and P1 and the difference between F z and Pz electrodes for each task using the same types of lines in the same time interval. There is a large decrease of activity for the MEM task that becomes prominent just before the onset of the eye movement.

mentary motor area (SMA), the posterior parietal cortex (PPC), the superior temporal cortex (ST), the cingulate cortex and the basal ganglia (BG) [24,31]. The larger activity in frontal sites compared to the parietal sites found in our experiment is in accordance with the findings of previous ERP studies. In a hand pointing delay paradigm Geffen et al. [14] reported a frontal activation during the memory condition. Furthermore a frontal ERP activation was the main finding in a letter memory paradigm in which subjects had to remember increasing numbers of letters. This frontal activation increased with increasing memory load [28]. A frontal activation during the delay period was also found in a remembered saccade task [27]. Thus, the frontal activation was a constant finding related to the condition in which subjects had to postpone action and

maintain information in working memory. In our study we observed a pattern of larger frontal negativity compared to the parietal activity for the memory task only during the early delay period. Thus memory-related EEG changes were not restricted to a frontal activation but encompassed a concomitant participation of the parietal areas in a more complex pattern of activation. Since we did not use source analysis due to the limited number of electrode sites recorded, we were not at present in the position to speculate further as to the exact anatomical locations of the sources responsible for this pattern of activation. This memory-related differential activity disappeared during the late delay period although a pattern of larger frontal versus parietal activity was observed when activity for all tasks was put together. This lack of a specific effect of the

I. Evdokimidis et al. / Cognitive Brain Research 12 (2001) 89 – 99

memory task could be due to the overall signal reduction with increasing delay that was observed in our study and was also observed in other ERP studies of working memory [14,28]. Ruchkin et al. [28] also reported in their study that the specific memory-related component of the ERP was lost at the end of the delay. A possible explanation for this loss of memory-related activity with delay might be that this activity becomes more restricted to particular areas as the delay increases and thus can not be detected using scalp recorded ERPs.

4.2. The pre-saccadic ERP: further evidence for specific memory-related activity Several ERP studies have already shown that the generation of a saccade is preceded by a central and parietal negativity suggesting a movement-related activation of several cortical areas. The amplitude and the topography of this pre-movement activation was affected by the type of saccade performed [7], as well as by the duration of the task, the motivation of the subjects, etc. [8]. Thus, self-paced saccades were preceded by a larger parietal negativity compared to visually guided saccades [20] and anti-saccades showed a frontal activity not observed with the visually guided saccades [7]. In this study the execution of remembered saccades was preceded by a central and parietal negativity, as was the case for the visually guided delay saccades. In addition the presence or not, of the visual target during saccade preparation had a significant effect on the ERP activity. Thus when the visual target was present the parietal activity spread over the central and frontal areas (DEL task). On the contrary the absence of the visual target (MEM and M-DEL tasks) resulted in reduced ERP activity of the central and frontal areas but not of the parietal area. An investigation of the difference between frontal and parietal activity for each task revealed a pattern of decreased frontal compared to parietal activity only for the memory task. Thus the pattern of frontal versus parietal difference in activity that was observed for the remembered saccades during the early delay period seemed to inverse during the pre-saccadic period. The reversal of activity for the memory task occurs as the trial evolves in time from the memory period to the pre-saccadic period. This is then exactly the type of information in time where the ERP has a unique advantage over the functional imaging studies where the effects of different events within one trial of a few seconds cannot be dissociated. The next question is then what are the areas that are involved in this pattern of activity and the combination of high electrode density ERP with functional imaging could provide the answer. At present we could hypothesize that this activity pattern might reflect the interactions between frontal and parietal cortex. The frontal or / and parietal involvement in the generation of remembered saccades have been postulated by several earlier studies. In humans, lesions of both the frontal (FEF,

97

DLPFC, PFC) and the parietal (PPC) areas result in a decrease in the accuracy and an increase in the latency of the remembered saccades [26]. Furthermore, data from transcranial magnetic stimulation (TMS) experiments revealed an interesting interplay between frontal and parietal areas during the performance of remembered saccades. The inhibition of the frontal areas by TMS resulted in a decrease in the accuracy of the remembered saccade only if this stimulation was delivered at a narrow time window of 700–1500 ms after the disappearance of the cue target [22,23]. This decrease in accuracy was interpreted as an indication of working memory dysfunction due to the TMS. Our findings also showed a memory-related increase in frontal activity compared to the parietal one at the beginning of the memory period. In the same TMS studies the parietal stimulation produced a prolongation of the saccade latency, only when the stimulation was applied at 100 ms after the go signal (pre-saccadic period), indicating a relation of parietal areas with the initiation of the saccade. We also observed a parietal activation that was related to the initiation of the saccadic eye movement. That activation was observed for all three tasks. A frontalparietal interaction in spatial working memory functions was also observed in single neuron recordings from these areas in the monkey. Chafee and Goldman-Rakic [5] used a spatial working memory task and observed significant similarities of neuronal activity between the dorsolateral prefrontal (DLPF) and the posterior parietal (PPC) cortices. Thus in both areas different neuronal populations were activated at the appearance of the cue, the delay and the presaccadic period. In their study Quintana and Fuster [25] showed that single neurons in the parietal and dorsolateral prefrontal cortex where related to the memory of a spatial location although their percentage was larger in the parietal cortex. The authors suggest that the certainty of the decision to move was reflected in the activity of both the dorsolateral prefrontal and parietal neurons. These decision-related neurons increased their firing rate as the response approached in time. It could thus be hypothesized that in a delay visuomotor paradigm a prefrontal-parietal network is activated in order to accomplish the ‘perception-action’ integration and this assumption is also supported by the anatomical connections of the two areas [17]. Our data show a pattern of frontal versus parietal activity differences both during the delay period and before the execution of remembered saccades. Although we could not say at the present time whether this pattern reflects the neuronal activity of frontal and parietal areas our findings seem to be different for those of studies that used neuronal recording in the sense that we find differences in activation between frontal and parietal areas while the neuronal recordings show simultaneous increases of neuronal firing in both frontal and parietal areas during the memory and motor preparation periods. These differences could be interpreted in view of the fact that ERPs capture the activation in time of very large neuronal populations and

98

I. Evdokimidis et al. / Cognitive Brain Research 12 (2001) 89 – 99

thus ‘frontal’ and ‘parietal’ activities are reflecting an average outcome that is very different from single neuron activity sampled for a small population of neurons located in a strictly defined brain area. Another interesting possibility could be that humans differ from macaques in the functional properties of cortical areas involved in the control of eye movements. It has been shown for example that the frontal eye field of the macaque and the human differ in their topography, their microstructure and their anatomical connections with other brain areas [32].

4.3. Saccade response latency: motor set reduces time needed for motor programming The reduction in response latency before the execution of remembered saccades is a well-known phenomenon [33]. We also observed a similar reduction of response latency with remembered saccades. This reduction could not be attributed to the absence of the peripheral target at movement onset through a mechanism of disengagement of attention [9] because the difference in response time between the two delay tasks was not significant. More interestingly we observed a further reduction of response time with increasing delay in all three tasks. This phenomenon of a decrease in response latency with increasing delay has also been observed in a memory arm movement task [29]. Thus by using different motor outputs and different delay conditions (memory versus delay) we observe the same phenomenon of a reduction in response latency when an increase is observed in the interval between visual stimulus presentation and the signal for movement execution. This reduction could be an indication that a long motor set activation results in a reduction of the time required for specific motor programming. This hypothesis requires neurophysiological validation.

5. Conclusions We recorded ERPs in one memory and two delay tasks. Early on during the delay period a larger activity of frontal versus parietal areas was evident for the memory task only. Later in the delay this specific memory-related pattern of activity disappeared. After the go signal a parietal negativity was observed. This negativity was spread over the central and frontal areas only for the delay task. The absence of a visual target during saccade preparation resulted in a smaller central and frontal activity. A specific decrease in activity of frontal versus parietal areas was observed only in the memory task indicating a near reversal in time of the activity pattern observed during the early delay period. These findings stress the importance of time in the investigation of activation patterns during the execution of cognitive tasks. Thus the ERP could offer a more general view of the interactions in time of different brain areas and the relation of these interactions to

cognitive process such as working memory. These findings could then complement the findings from neuroimaging studies that offer a more precise localization of the areas involved.

Acknowledgements This work was supported by the National University of Athens Research Funding 70 / 4 / 3508. We would like to thank Dr. A. Balogh for helping us with editing the manuscript and Professor M. Dalakas for his ongoing support to our group.

References [1] R.A. Andersen, The role of inferior parietal lobule in spatial perception and visual-motor integration, in: F. Plum, V.B. Mountcastle, S.R. Geiger (Eds.), The Handbook of Physiology, Section 1: The Nervous System, Vol. V: Higher functions of the brain, Part 2, American Physiological Society, Bethesda, MD, 1987, pp. 483–518. [2] R.A. Andersen, R.M. Bracewell, S. Barash, J.W. Gnadt, L. Fogassi, Eye position effects on visual, memory and saccade-related activity in areas LIP and 7a of macaque, J. Neurosci. 10 (1990) 1176–1196. [3] S. Barash, R.M. Bracewell, L. Fogassi, J.W. Gnadt, R.A. Andersen, Saccade-related activity in the lateral intraparietal area. I. Temporal properties; comparison with area 7a, J. Neurophysiol. 66 (1991) 1095–1108. [4] S. Barash, R.M. Bracewell, L. Fogassi, J.W. Gnadt, R.A. Andersen, Saccade-related activity in the lateral intraparietal area. II. Spatial properties, J. Neurophysiol. 66 (1991) 1109–1124. [5] M.V. Chafee, P.S. Goldman-Rakic, Matching patterns of activity in primate prefrontal area 8a and parietal area 7ip neurons during a spatial working memory task, J. Neurophysiol. 79 (1998) 2919– 2940. [6] I. Evdokimidis, T. Mergner, C.H. Lucking, Dependence of presaccadic potentials on the type of the saccadic eye movements, Electroencephalogr. Clin. Neurophysiol. 83 (1992) 179–191. [7] I. Evdokimidis, C. Liakopoulos, T.S. Constantinidis, C. Papageorgiou, Cortical potentials with antisaccades, Electroencephalogr. Clin. Neurophysiol. 98 (1996) 377–384. [8] I. Evdokimidis, T.S. Constantinidis, P. Gourtzelidis, C. Liakopoulos, C. Papageorgiou, Changes of presaccadic cortical activity when performing horizontal, visually guided saccades, Electroencephalogr. Clin. Neurophysiol. 102 (1997) 256–260. [9] B. Fischer, B. Breitmeyer, Mechanisms of visual attention revealed by saccadic eye movements, Neuropsychologia 25 (1987) 73–83. [10] S. Funahashi, C.J. Bruce, P.S. Goldman-Rakic, Mnemonic coding of visual space in the monkey’s dorsolateral prefrontal cortex, J. Neurophysiol. 61 (1989) 331–349. [11] S. Funahashi, C.J. Bruce, P.S. Goldman-Rakic, Visuospatial coding in primate prefrontal neurons revealed by oculomotor paradigms, J. Neurophysiol. 63 (1990) 814–831. [12] S. Funahashi, C.J. Bruce, P.S. Goldman-Rakic, Neuronal activity related to saccadic eye movements in the monkey’s dorsolateral prefrontal cortex, J. Neurophysiol. 65 (1991) 1464–1483. [13] S. Funahashi, M.V. Chafee, P.S. Goldman-Rakic, Prefrontal neuronal activity in rhesus monkeys performing a delayed anti-saccade task, Nature 365 (1993) 753–756. [14] G.M. Geffen, M.J. Wright, H.J. Green, N.A. Gillespie, D.C. Smyth, D.M. Evans, L.B. Geffen, Effects of memory load and distraction on

I. Evdokimidis et al. / Cognitive Brain Research 12 (2001) 89 – 99

[15]

[16]

[17]

[18]

[19]

[20]

[21] [22]

[23]

[24]

performance and event-related slow potentials in a visuospatial working memory task, J. Cogn. Neurosci. 9 (1997) 743–757. J.W. Gnadt, R.A. Andersen, Memory related motor planning activity in posterior parietal cortex of macaque, Exp. Brain Res. 70 (1988) 216–220. J.W. Gnadt, L.E. Mays, Neurons in monkey parietal area LIP are tuned for eye-movement parameters in three-dimensional space, J. Neurophysiol. 73 (1995) 280–297. P.S. Goldman-Rakic, Topography of cognition: parallel distributed networks in primate association cortex, Annu. Rev. Neurosci. 11 (1988) 137–156. D. Guitton, H.A. Buchtel, D.A. Douglas, Frontal lobe lesions in man cause difficulties in suppressing reflexive glances and in generating goal-directed saccades, Exp. Brain Res. 58 (1985) 455–472. J. Jonides, E.E. Smith, R.A. Koeppe, R.A. Awh, S. Minoshima, M.A. Mintum, Spatial working memory in humans as revealed by PET, Nature 363 (1993) 623–625. D. Kurtzberg, H.G. Vaughan, Topographic analysis of human cortical potentials preceding self-initiated and visually triggered saccades, Brain Res. 243 (1982) 1–9. R.J. Leigh, D.S. Zee, in: The Neurology of Eye Movements, 2nd Edition, F.A. Davis, Philadelphia, PA, 1991, p. 83. ¨ S.I. Vermersch, S. Rivaud, B. Gaymard, C. PierrotR.M. Muri, Deseilligny, Effects of single-pulse transcranial magnetic stimulation over the prefrontal and posterior parietal cortices during memory-guided saccades, J. Neurophysiol. 76 (1996) 2102–2106. ¨ B. Gaymard, S. Rivaud, A.L. Vermersch, C. Hess, C. R.M. Muri, Pierrot-Deseilligny, Hemispheric asymmetry in cortical control of memory-guided saccades. A transcranial magnetic stimulation study, Neuropsychologia 38 (2000) 1105–1111. E.P. O’Sullivan, I.H. Jenkins, L. Henderson, C. Kennard, D.J.

[25]

[26]

[27]

[28]

[29]

[30] [31]

[32]

[33]

99

Brooks, The functional anatomy of remembered saccades: a PET study, Neuroreport 6 (1995) 2141–2144. J. Quintana, J.M. Fuster, From perception to action: temporal integrative functions of prefrontal and parietal neurons, Cereb. Cortex 9 (1999) 213–221. C. Pierrot-Deseilligny, S. Rivaud, B. Gaymard, Y. Agid, Cortical control of memory-guided saccades in man, Exp. Brain Res. 83 (1991) 607–617. ¨ ¨ S. Carlson, J. Kekoni, H. Hamalainen, ¨ ¨¨ P. Rama, A spatial oculomotor memory-task performance produces a task-related slow shift in human electroencephalography, Electroencephalogr. Clin. Neurophysiol. 94 (1995) 371–380. D.S. Ruchkin, R. Johnson Jr., H. Canoune, W. Ritter, Short-term memory storage and retention: an event-related brain potential study, Electroencephalogr. Clin. Neurophysiol. 76 (1990) 419–439. N. Smyrnis, P. Gourtzelidis, I. Evdokimidis, A systematic directional error in 2-D arm movements increases with increasing delay between visual target presentation and movement execution, Exp. Brain Res. 131 (2000) 111–120. L.H. Snyder, A.P. Batista, R.A. Andersen, Coding of intention in the posterior parietal cortex, Nature 386 (1997) 167–170. J.A. Sweeney, M.A. Mintun, S. Kwee, M.B. Wiseman, D.L. Brown, D.R. Rosenberg, J.R. Carl, Positron emission tomography study of voluntary saccadic eye movements and spatial working memory, J. Neurophysiol. 75 (1996) 454–468. E.J. Tehovnik, M.A. Sommer, I. Chou, W.M. Slocum, P.H. Schiller, Eye fields in the frontal lobes of primates, Brain Res. Rev. 32 (2000) 413–448. J.M. White, D.L. Sparks, T.R. Stanford, Saccades to remembered target locations: an analysis of systematic and variable errors, Vision Res. 34 (1993) 79–92.