Exp Brain Res (2001) 139:287–296 DOI 10.1007/s002210100765

R E S E A R C H A RT I C L E

Nikolaos Smyrnis · Dionisios Linardatos Ioannis Evdokimidis · Theodoros S. Constantinidis Costas N. Stefanis

An early transient 40 Hz activity discriminates a following pro-saccade from a no-move and anti-saccade choice Received: 17 October 2000 / Accepted: 23 March 2001 / Published online: 26 June 2001 © Springer-Verlag 2001

Abstract We studied the oscillatory activity of the scalp-recorded EEG in healthy humans performing a task that required a particular eye-movement response choice according to the shape of a visual target. We observed a significant stimulus-aligned activity at the 40 Hz frequency band 100 ms after the appearance of the target only when that target was the end point for the subsequent eye movement (pro-saccade). This activity was most prominent over the central-parietal area of the right hemisphere. When the target indicated a movement to the opposite direction (anti-saccade) or indicated that no movement was required (no-move), this 40 Hz activity was nearly absent. This difference in activity between the pro-saccade and the other two tasks was evident in the single subject ERPs for four of the six subjects studied. In contrast, the movement-aligned 40 Hz activity for the pro-saccade and anti-saccade was almost identical. We speculate that this early stimulus-aligned 40 Hz activity might reflect a fast transformation of a visual stimulus to a motor response (eye movement) that can be performed for the pro-saccade task where stimulusresponse compatibility is strong compared to the antisaccade and no-move tasks. The movement-aligned 40 Hz activity might be related to the motor response preparation per se. We conclude that this task specific transient oscillatory activity could be used as a probe in the study of the temporal dynamics of visuomotor transformations. N. Smyrnis (✉) · D. Linardatos · T.S. Constantinidis C.N. Stefanis University Mental Health Research Institute, National University of Athens, Athens, Greece e-mail: [email protected] Tel.: +30-1-7289115, Fax: +30-1-7216474 N. Smyrnis · I. Evdokimidis · T.S. Constantinidis Cognition and Action Group, Neurology Department, National University of Athens, Aeginition Hospital, 72Vas. Sofias Avenue, Athens 11528, Greece D. Linardatos Division of Communications and Signal Processing, Department of Informatics, National University of Athens, Athens, Greece

Keywords Gamma-band response · Oscillatory EEG activity · Visual discrimination · Oculomotor task · GO/NOGO · Anti-saccade · Visuomotor transformation

Introduction High-frequency oscillatory activity, also called 40 Hz activity or gamma-band response (GBR), has been observed in the frequency analysis of scalp-recorded event related electric potentials (ERPs) and magnetic fields (MEG) using auditory (Galambos et al. 1981; Pantev et al. 1991; Jokeit and Makeig 1994; Joliot et al. 1994), somatosensory (Desmedt and Tomberg 1994) and visual stimuli (Lutzenberger et al. 1995; Tallon-Baudry et al. 1996). Cognitive processes also elicit gamma-band activity, including selective auditory attention (Tiitinen et al. 1993), visuospatial attention (Gruber et al. 1999), coherent object representation (Tallon-Baudry et al. 1996), visual search (Tallon-Baudry et al. 1997), short term-memory (Tallon-Baudry et al. 1998) and perception of a gestalt (Keil et al. 1999). At the neurophysiological level, oscillatory activity in the 40 Hz range has been observed in single unit recordings of the anesthetized cat (Gray and Singer 1989; Gray et al. 1989, 1990, 1992) and alert monkey (Eckhorn et al. 1988; Krieter and Singer 1992; Frien et al. 1994) visual cortex. Singer and Gray (1995) proposed that the oscillatory activity of visual cortical neurons reflects the binding of information regarding different properties of the visual stimulus. It has been proposed that high-frequency oscillations in the EEG and MEG reflect the activation of thalamocortical circuits during processing of perceptual stimuli in the conscious state (Ribary et al. 1991; Llinas and Ribary 1993). Others have proposed that the scalp-recorded visual GBR might reflect the construction of a coherent object representation (Tallon-Baudry and Bertrand 1999). Thus 40 Hz activity can be elicited in different areas during different cognitive operations and it could provide specific information about these operations that is not available in the traditional analysis of the event re-

288

lated potentials. In a recent investigation of oscillatory activity preceding saccadic eye movements in humans, a difference in the alpha and beta bands of the EEG was observed between express and regular saccades (Skrandies and Anagnostou 1999). Both types of saccades elicited similar movement related gamma band activity. The visual stimulus related oscillatory activity was not investigated in that study. When a visual stimulus serves as the instruction for a motor response, a stream of processing from visual areas to parietal areas is engaged (Ungerleider and Mishkin 1982; Goodale and Milner 1992). The association of a visual stimulus with a desired response is associated with activity of neurons in the dorsal parietal cortex (Mountcastle et al. 1975; Andersen 1989). Furthermore, recent work has suggested complex functions for this area such as the coding of the intention to move (Andersen et al. 1997; Snyder et al. 1997) and of decision related processing (Platt and Glimcher 1999). In previous work, we recorded ERPs over frontal and parietal areas during three variants of an oculomotor task in which subjects had to decide, according to the shape of the target, to make a saccade to a visual target, a saccade in the opposite direction of the target, or to withhold their response (Evdokimidis et al. 1996). The event related potentials (ERPs) showed differences among tasks in the late stages of processing close to the movement onset. In this study, we analyzed the high-frequency oscillatory activity (25–50 Hz) embedded in the ERP that was elicited in each task in order to investigate a possible role of highfrequency EEG components in the target discrimination, response selection and motor planning processes involved in these tasks.

Materials and methods We used the EEG data recordings from a previous study. The experimental protocol was approved by the scientific and ethics committee of Aeginition Hospital. The recording procedure and

Fig. 1 Response latency distributions. This figure presents the latency distributions (A pro-saccade, B anti-saccade). The vertical line represents the time period of 350 ms after stimulus onset that we retained for analysis of our EEG data (see Materials and methods)

the configuration of the stimuli are discussed in detail in the previous study (Evdokimidis et al. 1996). Each trial started with the appearance of a central fixation target (blue circle, 0,5 degrees of visual angle) on a computer monitor. The central target was extinguished after 6–8 s and, after a variable period of 100–400 ms, a peripheral target appeared at 10 degrees to the right or to the left of the central target. If the peripheral target was a circle, the subject was instructed to saccade on the target (pro-saccade); if the target was a square, the subject was instructed to make a saccade on the opposite side of the target (anti-saccade); and finally, if the target was a triangle, the subject had to continue fixating at the position of the central target (no-move). The peripheral targets were of equal size (0,5 degrees of visual angle) and color (yellow). Only data from six subjects out of ten were successfully retrieved from the original data set due to a technical problem. The total number of trials for each recording site was 1439, consisting of 446 pro-saccade trials, 462 anti-saccade trials and 531 no-move trials. The EEG was recorded from 25 electrode sites using the modified 10/20 system (F3, F1, Fz, F2, F4, FC3, FC1, FCz, FC2, FC4, C3, C1, Cz, C2, C4, CP3, CP1, CPz, CP2, CP4, P3, P1, Pz, P2, P4). The signal was band pass filtered (0.01–75 Hz) and digitized at 200 Hz. The horizontal and vertical movements of the right eye were recorded (EOG; DC recording) and digitized at 200 Hz. All trials with artifacts (blinks etc.) or eye movements that did not conform with the given instruction were rejected from further analysis. The EEG record retained for analysis for the pro-saccade and anti-saccade trials consisted of the time between the onset of the peripheral target and the onset of the eye movement, while the record retained for analysis for the no-move trial consisted of 350 ms after the onset of the corresponding peripheral target. Two time periods were analyzed: the 350 ms following presentation of the peripheral target (stimulus-aligned) and the 350 ms preceding onset of eye movement (movement-aligned). This time interval for analysis was chosen because approximately 99% of trials for the pro-saccade and anti-saccade task had response times equal or larger than this interval (Fig. 1). For each trial, the mean value corresponding to 50 ms before the onset of the peripheral target was used as a baseline and subtracted from all EEG values. The single trial data were then averaged to obtain a stimulus-aligned and movement-aligned ERP for each subject, for each task and each electrode site (25 electrode sites as described in Evdokimidis et al. 1996). Further averaging of individual subject averages led to stimulus-aligned and movement-aligned grand average ERPs for each task at each electrode site. The oscillatory activity was obtained using a wavelet decomposition of the stimulus-aligned and movement-aligned average

289 ERP for each subject and each task and of the stimulus-aligned and movement-aligned grand average ERP for each task. The Wavelet Packet Analysis (WPA) (Akay 1998), a generalization of the wavelet decomposition that re-analyzes the resulting signal details on various frequency bands, was used: WPA( j, n, k ) = ∑ x(t )Wj,n,k (t )

(1)

t

where x(t) is the ERP record, Wj,n,k(t) represents the analyzing functions, j is the scale parameter, n is the frequency parameter (n=0,1..15) and k is the time-localization parameter. The analyzing functions are defined by: Wj,n,k(t)=2–j/2Wn(2–jt–k)

(2)

where the set Wj,n=(Wj,n,k(t), k∈Z) is the (j,n) wavelet packet. Function Wn(t) is recursively defined as W2n (t ) = 2 ∑ h(k )Wn (2t − k ), W2n+1(t ) = 2 ∑ g(k )Wn (2t − k ) (3) k ∈Z

k ∈Z

where h(k) and g(k) are the reversed versions of the low-pass decomposition filter and the high-pass decomposition filter, respectively. Note that W0(t)=φ(t) is the scaling function and W1(t)=ψ(t) is the wavelet function. Computations were implemented with Matlab (Misiti et al. 1996) using the “sym2” wavelet function. The resulting individual details of the wavelet analysis were computed for the following bands of interest: a) 25–31.25 Hz, b) 31.25–37.5 Hz, c) 37.5–43.75 Hz and d) 43.75–50 Hz. The nonlinear energy operator (Maragos et al. 1991) for discrete-time signals ψd≡x2(t)–x(t–1)x(t+1)

(4)

was applied to these time series and the instantaneous energy signal was obtained. To test the significance of the 40 Hz activity (from here on defined as the oscillatory activity for the frequency band of 37.5–43.75 Hz) differences observed among tasks for the stimulus-aligned data, we used a random permutation procedure that does not rely on specific distribution assumptions for the data (Feinstein 1977). We reassigned the data for each trial of the original data set to a new task in a random fashion resulting in a new data set of shuffled trials. The shuffling process was performed on the complete data set of all trials for all subjects for each electrode site separately. The shuffled data were used to compute new grand average ERPs and then 40 Hz responses for each task and each electrode site. A set of ERP and 40 Hz activity values was retained for each one of 15 time windows along the time series of the grand average waveforms (15 time windows of 25 ms duration centered at 0, 25, 50... 350 ms after peripheral target onset). The differences between pro-saccade and anti-saccade, pro-saccade and no-move and anti-saccade and no-move for each time window and electrode site were computed. The permutation process was performed 2500 times resulting in 2500 shuffled differences for the stimulusaligned ERP and another 2500 differences for the stimulus-aligned 40 Hz activity. These differences were used to compute the percentage of times (transformed into a probability value from 0 to 1) that the shuffled ERP or 40 Hz activity difference between a pair of tasks was larger than the original difference for each time window and electrode site (differences in absolute value). If all the possible permutations among tasks could be performed, this percentage would reflect the exact two-tail probability that the original difference in energy could be obtained by a chance assignment of a trial to a particular task (Feinstein 1977). Figure 2 shows the probability (y-axis) that the absolute 40 Hz activity difference between the pro-saccade and the no-move tasks at the CP4 electrode site after N permutations (x-axis) was larger than the difference observed for the permutated data. It can be seen that after approximately 500 permutations the probability was <0.01and after 1000 permutations the probability practically did not change. For the purposes of our study we set the significance level to the probability of 0.01. Note that the in order to exhaust all possible permutations for the single trial data set, a number of 10681 permutations should be performed.

Fig. 2 Permutation results. This plot shows the probability (yaxis) that the absolute 40 Hz activity difference between the prosaccade and the no-move tasks after N permutations (x-axis) is larger than the original difference for the grand average 40 Hz activity for the CP4 electrode site. It can be seen that after 500 permutations, this probability becomes smaller than 0.01 and after 1000 permutations it is stable

In order to confirm our results from the permutation tests, we used the Kruskal-Wallis non-parametric ANOVA to test whether the ERP amplitude or the 40 Hz activity at each one of six time windows was significantly different among the three tasks. We used the amplitude or energy values at each one of the 50, 75, 100, 125, 150 and 175 ms time windows for each subject and each task for a set of 6 electrode sites (C4, C6 CP3, CPz, CP4, CP6) in which the permutation tests showed significant task differences. The energy or ERP amplitude at each time window was computed as for the permutation analysis by taking the average of five values (25 ms) centered at the particular time window value (for example 100 ms=average from 90–110 ms). We thus obtained a data set of 36 amplitude or energy values for each time window and each task (six subjects×six electrodes). The ANOVA for each time window was performed using the total of 108 (36×3) values of energy or ERP amplitude for the three tasks and using the task as the independent variable.

Results Stimulus-aligned oscillatory activity Figure 3 illustrates the grand average stimulus-aligned oscillatory activity at 40 Hz (defined as the 37.5– 43.75 Hz activity, see methods) for each of the three tasks at each of the 25 electrode sites (solid peaks). The stimulus-aligned grand average 40 Hz activity shows a clear dissociation between the pro-saccade task (Fig. 3A) and the other two tasks (Fig. 3B, C). A transient peak at 100 ms after stimulus presentation is clear and prominent for the pro-saccade task (Fig. 3A). This peak is much closer to the noise level of the baseline for the antisaccade task (Fig. 3B) and is negligible for the no-move task (Fig. 3C). The striking difference among tasks that was observed in the 40 Hz frequency band (37.5– 43.75 Hz), peaking at 100 ms after peripheral target onset, was not evident in the other frequency bands that were used to decompose the stimulus-aligned grand average ERP signal (Fig. 4) so we concentrated in our frequency analysis at this particular narrow frequency band. Figure 3 also shows the stimulus-aligned grand average ERPs, recorded for all three tasks (dotted lines). A clear ERP response peaking at 125 ms after the onset

290 Fig. 3 Stimulus-aligned 40 Hz and ERPs. Depicted are the grand average ERPs (dotted lines) and 40 Hz activity (solid peaks) for all electrode sites for which we recorded EEG data. The ERP waveforms are plotted in voltage (µV) and the 40 Hz activity is plotted in energy values (µV2). The total time after the onset of the peripheral target (marked with the vertical line) is 350 ms. A Prosaccade, B anti-saccade and C no-move

291

of the peripheral target can be observed for all three tasks. In conclusion we observed a clear dissociation among the three tasks at the 40 Hz frequency band. Figure 5 presents the 40 Hz activity extracted from the stimulus-aligned ERP for each subject for electrode site C2. A clear dissociation between the 40 Hz activity for the pro-saccade and the other two tasks is observed at 100 ms after stimulus onset in four out of the six subjects. In one subject (S5), the 40 Hz activity for the prosaccade and anti-saccade tasks was identical and larger than the activity for the no-move task while for subject S6 the activity for the anti-saccade task was larger than both the pro-saccade and the no-move tasks. The records of these last two subjects showed a low overall energy intensity at 100 ms after stimulus onset. A final observation was that in all subjects the activity at 100 ms for the no-move task was always the lowest. Permutation analysis confirmed a significant difference in 40 Hz activity between the pro-saccade and the anti-saccade as well between the pro-saccade and nomove tasks. As expected, the 40 Hz activity difference between the anti-saccade and no-move tasks was not significant. Figure 6 shows a map of gray scale coded probabilities that the difference of the permutated data between tasks was larger than the original difference (A, pro-saccade versus anti-saccade; B, pro-saccade versus no-move; C, anti-saccade versus no-move) for all electrode sites for each one of six time windows (50 –175 ms). Significant differences (P<0.01) between the pro-saccade and anti-saccade tasks appeared at the 50 ms window, were most pronounced at the 100 ms time window and were still present at the 125 ms and 75–150 ms windows. The large difference at 100 ms occurred primarily at the C2, C4 CP1, CPz, CP2 and CP4 electrode sites. The differences between the pro-saccade and no-move tasks were also significant for the 50–150 time windows and at even more electrode sites, whereas the difference between the anti-saccade and the no-move tasks was not significant. Permutation analysis did not show any significant task differences among the ERPs for the 50–175 ms time windows at any electrode sites. In summary, the permutation analysis confirmed that significant differences in the 40 Hz activity between the pro-saccade task and the other two tasks occurred around 100 ms after the onset of the peripheral target and had a central-parietal distribution with a dominance of the right hemisphere. Fig. 4 All high-frequency bands. Figure shows the grand average oscillatory activity for the three tasks (solid line pro-saccade, thin line anti-saccade, dotted line no-move) for three electrode sites in each frequency band: A for the 25–31.25 Hz band, B for the 31.25–37.5 Hz band, C for the 37.5–43.75 Hz band and D for the 43.75–50 Hz band. The oscillatory activity is plotted in energy values (µV2). A clear dissociation among tasks is evident at the 37.5–43.75 Hz band (named 40 Hz activity in the text). The activity peak at 100 ms for the pro-saccade task is larger than that for the anti-saccade and the no-move tasks. The same dissociation appears also at the 43.75–50 Hz band but the overall activity is smaller at this frequency. In the lower frequency bands the activity at 100 ms is very small and there are no task differences

292 Table 1 ANOVA for task differences. The results of the KruskalWallis ANOVA for each time window are depicted (H values and P level) for the subject 40 Hz activity (first column) and the ERPs (second column) for a specific set of electrode sites (see Materials and methods) Time window

GBR

ERP

75 ms 100 ms 125 ms 150 ms 175 ms

H=18.369, P<0.001 H=26.498, P<0.0001 H=25.624, P<0.0001 H=28.792, P<0.0001 H=4.919, NS

H=1.265, NS H=0.394, NS H=5.122, NS H=3.375, NS H=6.055, P=0.048

We confirmed the permutation results by performing Kruskal-Wallis ANOVAs at the 100–150 ms time windows using the individual 40 Hz energy values for each subject and each task, and, in a separate analysis, the ERP amplitude values for each subject and each task at the six electrode sites where the permutation showed the most significant task differences in 40 Hz activity (see methods). The results of this analysis are presented in Table 1. There were significant task differences in the 40 Hz activity but not the ERP for the 75–150 ms time windows, whereas there were no significant differences in the 40 Hz activity at 175 ms and a marginally significant difference in the ERPs. Movement-aligned 40 Hz activity

Fig. 5 Individual 40 Hz activity. This figure presents the 40 Hz activity computed from each subject's average ERP at the C2 electrode site. The 40 Hz activity is plotted in energy values (µV2). For subjects S1 to S4 a clear pattern of increased activation at 100 ms after stimulus onset (gray strip) was observed for the prosaccade task (solid line) compared with the anti-saccade task (thin line) and the no-move task (dotted line). For subject S5 the activity for the pro-saccade and the anti-saccade tasks was almost identical and larger than the activity for the no-move task while in subject S6 the activity for the anti-saccade was larger than the activity for the pro-saccade and the no-move). The activity at 100 ms for subjects S5 and S6 was lower than that for the other four subjects

We also present the movement-aligned 40 Hz activity for the pro-saccade and anti-saccade tasks in Fig. 7 (thin line: pro-saccade, thick line: anti-saccade). The activity was almost identical for the two tasks. The same effect was also observed for the oscillatory activity in all other frequency bands. Thus, the previously identified difference between the movement-aligned ERPs for the prosaccade versus anti-saccade tasks (Evdokimidis et al. 1996) was not observed for the oscillatory activity of the ERP. The movement-aligned 40 Hz activity appeared very late (the last 100 ms before movement onset) and peaked at 50 ms before the onset of the movement. It was most prominent at central-parietal electrode sites, as were the pre-movement ERPs recorded before saccadic eye movements (Evdokimidis et al. 1996). The last component of the 40 Hz activity measured 25 ms before movement onset overlapped movement onset. Thus high frequency oscillatory activity was in fact present prior to saccade initiation but, in contrast to the stimulus-aligned 40 Hz activity, the movement-aligned oscillatory activity did not dissociate the two tasks for any frequency band.

Discussion Significant task differences were observed for the stimulus-aligned 40 Hz activity at 100 ms after the appearance

293

Fig 6 Task differences. This series of plots is a graphical representation of the probability values (P values) that the task differences in 40 Hz activity resulting from the permutation analysis for each time window (labeled columns) were larger than the original task difference (see Materials and methods). Each plot presents a map of these probabilities for the 25 electrode sites using a contour mapping procedure for better visualization. A Pro-saccade versus anti-saccade, B pro-saccade versus no-move, C antisaccade versus no-move. It can be seen that a significant difference (P<0.01) started even at 50 ms after the appearance of the peripheral target but became clear at 100 ms. This significant difference was evident in series A and was more prominent in series B. The third series (C) showed no significant differences (P<0.01)

of the peripheral target. The pro-saccade task was the only task that elicited a robust 40 Hz activity following peripheral target onset. This task specific effect was only observed in the 40 Hz frequency band and was not evident in other high frequency bands and for the stimulusaligned ERP response. In addition, there was no difference in the movement-aligned 40 Hz activity between the pro-saccade and anti-saccade tasks. These stimulusaligned 40 Hz activity at 100 ms after stimulus onset observed for the pro-saccade task had a central-parietal distribution lateralized to the right hemisphere. A gamma-band activity 100 ms after stimulus onset has been observed in both visual (Buchner et al. 1999) and auditory tasks (Pantev et al. 1991; Tiitinen et al. 1993), and several hypotheses have been put forth re-

garding its site of origin, induction and modulation. In their work on the visually evoked gamma-band activity, Tallon-Baudry et al. (1996) have speculated that the phase-locked component of the response appearing at 100 ms after stimulus onset might result from a more general process of thalamocortical activation in accordance with the view of Llinas and Ribary (1993), given that this activity was not modulated by the visual stimulus properties. In addition, in a recent study Buchner et al. (1999) described two maxima for the visually induced phase-locked gamma-band activity at 100 ms after stimulus onset, one over the central electrode sites (Cz) and one over the occipital sites (Oz). The authors proposed that these maxima reflect the activity of two generators one originating at deep structures (thalamus) and one originating at the visual cortex. The stimulus-aligned 40 Hz activity we observed was specific to the pro-saccade task, although similar visual stimulation occurred in all three tasks and the ERP to the visual stimulus was similar for all three tasks both in magnitude and in latency. Furthermore the location of the maximum for this 40 Hz activity was central-parietal with right hemisphere predominance in our study. Thus, a general perceptual mechanism does not explain the task specificity of the stimulus-aligned 40 Hz activity. The acoustically evoked gamma-band response has been proposed to reflect activation of the auditory association cortex (Pantev et al. 1991). This auditory gamma-

294 Fig. 7 Movement-aligned 40 Hz activity. The grand average 40 Hz activity for the prosaccade task (thick lines) and the anti-saccade task (thin lines) is depicted for the 25 electrode sites. The 40 Hz activity is plotted in energy values (µV2). The activity is aligned to the movement onset (far right vertical lines) and the total time is 350 ms. The 40 Hz activity for the Cz electrode is re-plotted and enlarged in the bottom center of the figure

band activity was modulated by attention (Tiitinen et al. 1993). We cannot attribute the stimulus-aligned 40 Hz activity to selective attention mechanisms because in all three tasks the subjects were required to attend to all three types of targets in order to make the appropriate response. In addition, and perhaps more importantly, the subjects' performance in all three tasks was almost without error (see Evdokimidis et al. 1996) and we measured the oscillatory activity of the ERP averaged for correct trials only. Thus attention mechanisms cannot explain our results either. Specific task differences must account for the task specificity of the stimulus-aligned 40 Hz activity. In our pro-saccade task, the visual stimulus served as the eye movement target thus requiring a visuomotor transformation. In the no-move task no such transformation was required while in the anti-saccade task the direction information was required to plan a saccade at the opposite direction. We could postulate then that a specific relation exists between visuomotor transformation and the stimulus-aligned 40 Hz activity. The right central-parietal maximum of this response may be a further indication of visuospatial processing given the well established role played by the right parietal hemisphere (Bisiach and Vallar 1988) in such processes. Furthermore, Andersen et al. (1997) have postulated that the posterior parietal cortex is suitably interfaced between visual and motor

areas to serve as a specific module for visuomotor transformations relevant to upcoming eye or arm movements (Gnadt and Andersen 1988; Mazzoni et al. 1996; Andersen et al. 1997). In addition, the same group showed that when attention to particular location is dissociated from intention to move, activity of posterior parietal neurons primarily reflects the intention to move (Snyder et al. 1997). Our results also indicate that a mechanism of spatial attention to the target location (Corbetta et al. 1993) cannot explain the stimulus-aligned 40 Hz activity for the pro-saccade task only. A motor intention though could be present in all three tasks, activating the parietal cortex which would then inhibited by the activation of frontal cortex in the case of an anti-saccade (Guitton et al. 1985) or a no-move condition (Kawashima et al. 1996). In a recent study, posterior parietal cortical neurons were robustly activated before the execution of antisaccades (Gottlieb and Goldberg 1999). Thus a visuomotor transformation could be present in all three tasks but the timing of this transformation and the participation of the posterior parietal cortex in it, could vary from task to task. If subjects could learn after the first few trials a “rule” of associating the curved target (circle) with the pro-saccade (the other two target shapes have corners) they could search for curvature in the shape of the target and make a very fast decision leading to a direct computation of a motor intention in the case of the pro-saccade

295

task. In the case of the other two tasks where the stimulus-response compatibility was not strong subjects could use a different strategy of delaying their decision until the visual discrimination of the target was completed. This early discrimination and/or motor intention computation might then be related to the appearance of the stimulus locked 40 Hz response. Indications from visual identification paradigms suggest that visual discriminations occur very early on in the processing of visual stimuli (Rolls and Tovee 1994). Furthermore, Gnadt and Andersen (1988) have shown in a memory saccade paradigm that the direction specific activity of neurons in the posterior parietal cortex appears less than 100 ms after stimulus onset. Thus, both a visual discrimination and visuomotor transformation might occur very early (100 ms) after the appearance of the stimulus. The hypothesis that the transient 40 Hz activity might reflect a fast visuomotor transformation in the parietal cortex when stimulus-response compatibility is strong could also explain the results of a study that measured the gamma-band activity in a GO/NOGO task (Shibata et al. 1999) in which two visual stimuli (arrowheads) were used that instructed subjects either to press a button (GO condition) or withhold their response (NOGO condition). Spectral analysis of the evoked response showed no significant activity in the 40 Hz frequency range for either of the task conditions. Given that the stimulus to GO (arrowhead pointing down) was not the spatial target for the manual response (key press) the stimulus-response compatibility was not large and thus precluded a fast visuomotor transformation that would be associated with an early 40 Hz response. A remaining issue though is the fact that although the stimulus-aligned 40 Hz activity appearing at 100 ms after stimulus onset discriminates between the pro-saccade and the other two tasks there is a considerable delay (200 ms) between the activation in response to the hypothesized visual discrimination and/or motor planning and the final threshold for a motor response. A significant decrease in response latency was observed for the pro-saccade task compared to the anti-saccade task but it was very small (difference of 26.5 ms, t=–4.31, P<0.01, see Fig. 1). We observed a task difference only for the stimulusaligned 40 Hz activity. The movement-aligned 40 Hz activity for the pro-saccade and anti-saccade tasks was almost identical. A clear peak was observed 50 ms before the onset of the eye movement and was also robust just before the onset of the response (25 ms before the movement onset). This late component might be related to a muscular artifact called the “spike potential” (Thickbroom and Mastaglia 1985). Movement related gamma band activity has also been observed just before the execution of finger movements (Pfurtscheller et al. 1994). The spatial distribution of the movement-aligned 40 Hz activity was widespread with greatest activity at central and parietal areas, similar to the distribution of the movement-aligned ERP recorded before visually triggered saccades (Evdokimidis et al. 1992). This activity,

indistinguishable for the pro-saccade and anti-saccade tasks, might be related to the process of movement preparation. In conclusion, a clear 40 Hz activity peak at 100 ms after stimulus onset occurred only when the stimulus was the target of the upcoming eye movement (pro-saccade). The maximum of this activity occurred over the centralparietal area of the right hemisphere. It is postulated that this signal might reflect the early processing at the posterior parietal cortex of a transformation of a visual stimulus to a motor response. The specificity of this activity both in terms of time (after stimulus onset) and task (pro-saccade) although not fully explained in terms of the behavioral outcome is interesting and points to further exploration of this signal and its significance as a probe in the study of fast visuomotor transformations. Acknowledgements This work was supported by the 97/125/EL Program of the General Secretary for Research and Technology of Greece awarded to Costas N. Stefanis. We would like to thank Professor Apostolos P. Georgopoulos for helpful advice in the data analysis and Dr. Allison Balogh for editing the manuscript.

References Akay M (1998) Time frequency and wavelets in biomedical signal processing. IEEE Press, New York Andersen R (1989) Visual and eye movement functions of the posterior parietal cortex. Annu Rev Neurosci 12:377–403 Andersen R, Snyder LH, Bradley DC, Xing J (1997) Multimodal representation of space in the posterior parietal cortex and its use in planning movements. Annu Rev Neurosci 20:303–330 Bisiach E, Vallar G (1988) Hemineglect in humans. In: Boller F, Grafman J (eds) Handbook of neuropsychology. Elsevier, Amsterdam, pp 195–222 Buchner H, Gobbele R, Waberki TD, Wagner M, Fuchs M (1999) Evidence for independent thalamic and cortical sources involved in the generation of the visual 40 Hz response in humans. Neurosci Lett 269:59–62 Corbetta M, Miezin FM, Shulman GL, Petersen SE (1993) A PET study of visuospatial attention. J Neurosci 13:1201–1226 Desmedt JE, Tomberg C (1994) Transient phase-locking of 40 Hz electrical oscillations in prefrontal and parietal human cortex reflects the process of conscious somatic perception. Neurosci Lett 168:126–129 Eckhorn R, Bauer R, Jordan W, Brosch M, Kruse W, Reitboeck HJ (1988) Coherent oscillations: a mechanism of feature linking in the visual cortex? Biol Cybern 60:121–130 Evdokimidis I, Mergner T, Lücking CH (1992) Dependence of presaccadic cortical potentials on the type of saccadic eye movements. Electroencephalogr Clin Neurophysiol 83:179– 191 Evdokimidis I, Liakopoulos D, Constantinidis TS, Papageorgiou C (1996) Cortical potentials with antisaccades. Electroencephalogr Clin Neurophysiol 98:377–384 Feinstein AR (1977) Permutation tests and statistical significance. In Clinical biostatistics. C.V. Mosby, Saint Louis, pp 287–304 Frien A, Eckhorn R, Bauer R, Woelbern T, Kehr H (1994) Stimulus-specific fast oscillations at zero phase between visual areas V1 and V2 of awake monkey. Neuroreport 5:2273–2277 Galambos R, Makeig S, Talmachoff PJ (1981) A 40-Hz auditory potential recorded from the human scalp. Proc Natl Acad Sci USA 86:1698–1702 Gnadt JW, Andersen RA (1988) Memory related motor activity in posterior parietal cortex of the macaque. Exp Brain Res 70:216–220

296 Goodale MA, Milner DA (1992) Separate visual pathways for perception and action. Trends Neurosci 15:20–25 Gottlieb J, Goldberg ME (1999) Activity of neurons in the lateral intraparietal area of the monkey during an antisaccade task. Nat Neurosci 2:906–912 Gray CM, Singer W (1989) Stimulus specific neuronal oscillations in orientation columns of cat visual cortex Proc Natl Acad Sci USA 86:1698–1702 Gray CM, König P, Engel AK, Singer W (1989) Oscillatory responses in cat visual cortex exhibit inter-columnar synchronization which reflects global stimulus properties. Nature 338: 334–337 Gray CM, Engel AK, König P, Singer W (1990) Stimulus-dependent neuronal oscillations in cat visual cortex: receptive fields properties and feature dependence. Eur J Neurosci 2:607–619 Gray CM, Engel AK, König P, Singer W (1992) Synchronization of oscillatory neuronal responses in cat striate cortex: temporal properties. Vis Neurosci 8:337–347 Gruber T, Müller MM,Keil A, Elbert T (1999) Selective visualspatial attention alters induced gamma band responses in the human EEG. Clin Neurophysiol 110:2074–2085 Guitton D, Buchtel HA, Douglas DA (1985) Frontal lobe lesions in man cause difficulties in suppressing reflexive glances and in generating goal-directed saccades Exp Brain Res 58:455– 472 Jokeit H, Makeig S (1994) Different event-related patterns of γ-band power in brain waves of fast- and slow-reacting subjects. Proc Natl Acad Sci USA 91:6339–6343 Joliot M, Ribary U, Llinas R (1994) Human oscillatory brain activity near 40 Hz coexists with cognitive temporal binding. Proc Natl Acad Sci USA 91:11748–11751 Kawashima R, Kazunori S, Hiroshi I, Shuichi O, Furumoto S, Gotoh R, Koyama M, Yoshioka S, Takahashi T, Takahashi K, Yanagisawa T, Fukuda H. (1996) Functional anatomy of GO/NOGO discrimination and response selection – a PET study in man. Brain Res 728:79–89 Keil A, Müller MM, Ray WJ, Gruber T, Elbert T (1999) Human gamma band activity and the perception of a Gestald. J Neurosci 19:7152–7161 Krieter AK, Singer W (1992) Oscillatory neuronal responses in the visual cortex of the awake macaque monkey. Eur J Neurosci 4:369–375 Llinas R, Ribary U (1993) Coherent 40-Hz oscillations characterizes dream state in human. Proc Natl Acad Sci USA 90: 2078–2081 Lutzenberger W, Pulvermüller F, Elbert T, Birbaumer N (1995) Visual stimulation alters local 40 Hz responses in humans: an EEG study. Neurosci Lett 183:39–42 Maragos P, Kaiser JF, Quatieri TF (1991) Energy separation in signal modulations with application to speech analysis. IEEE Trans Signal Proc 41:3024–3051 Mazzoni P, Bracewell RM, Barash S, Andersen RA (1996) Spatially tuned auditory responses in area LIP of macaques performing delayed memory saccades to acoustic targets. J Neurophysiol 75:1233–1241

Misiti M, Misiti Y, Oppenheim G, Poggi JM (1996) Wavelet Toolbox: for use with MATLAB, Wavelet Toolbox User's Guide Mountcastle VB, Lynch JC, Georgopoulos A, Sakata H, Acuna C (1975) Posterior parietal association cortex of the monkey: command functions for operations within extrapersonal space. J Neurophysiol 38:871–908 Pantev C, Makeig S, Hoke M, Galambos R, Hampson S, Gallen C (1991) Human auditory evoked gamma-band magnetic fields. Proc Natl Acad Sci USA 88:8996–9000 Pfurtscheller G, Flotzinger D, Neuper C (1994) Differentiation between finger, toe and tongue movement in man based on 40 Hz EEG. Electroencephalogr Clin Neurophysiol 90:456– 460 Platt ML, Glimcher PW (1999) Neural correlates of decision variables in parietal cortex. Nature 400:233–238 Ribary U, Ioannides AA, Singh KD, Hasson R, Bolton JPR, Lado F, Mogilner A, Llinas R (1991) Magnetic tomography of coherent thalamocortical 40-Hz oscillations in humans. Proc Natl Acad Sci USA 88:11037–11041 Rolls ET, Tovee MJ (1994) Processing speed in the cerebral cortex and the neurophysiology of visual masking Proc R Soc Lond B 257:9–15 Shibata T, Shimoyama I, Ito T, Abla D, Iwasa H, Koseki K, Yamanouchi N, Sato T, Nakajima Y (1999) Event-related dynamics of the gamma-band oscillation in the human brain: information processing during a GO/NOGO hand movement task. Neurosci Res 33:215–222 Singer W, Gray CM (1995) Visual feature integration and the temporal correlation hypothesis. Annu Rev Neurosci 18:555–586 Skrandies W, Anagnostou E (1999) Electroencephalographic cortical oscillations and saccadic eye movements in humans. Neurosci Lett 261:57–60 Snyder LH, Batista AP, Andersen RA (1997) Coding of intention in the posterior parietal cortex. Science 386:167–170 Tallon-Baudry C, Bertrand O (1999) Oscillatory gamma activity in humans and its role in object representation. Trends Cognit Sci 3:151–158 Tallon-Baudry C, Bertrand O, Deluech C, Pernier J (1996) Stimulus specificity of the phase-locked and non-phase-locked 40 Hz visual responses in human. J Neurosci 16:4240–4249 Tallon-Baudry C, Bertrand O,Deluech C, Pernier J (1997) Oscillatory γ-band (30–70 Hz) activity induced by a visual search task in humans. J Neurosci 17:722–734 Tallon-Baudry C, Bertrand O, Peronnet F, Pernier J (1998) Induced γ-band activity during the delay of a visual short-term memory task in humans. J Neurosci 18:4244–4254 Thickbroom GW, Mastaglia FL (1985) Presaccadic “spike” potential: investigation of topography and source. Brain Res 339: 271–280 Tiitinen H, Sinkkonen K, Reinikainen K, Alho k, Lavikainen J, Näätänen R (1993) Selective attention enhances the auditory 40-Hz transient response in humans. Nature 364:59–60 Ungerleider LG, Mishkin M (1982) Two cortical visual systems In: Ingle DJ, Goodale MA, Mansfield RJW (eds) Analysis of visual behavior. MIT Press, Cambridge, pp 549–586