Brain, Vol. 122, No. 5, 915-931,
May 1999
© 1999 Oxford University Press
Cognitive motor control in human pre-supplementary motor area studied by subdural recording of discrimination/selection-related potentials
1 Departments of Brain Pathophysiology, 2 Neurosurgery and 3 Neurology, Kyoto University School of Medicine,Kyoto, Japan
Correspondence to:
Akio Ikeda MD, Department of Brain Pathophysiology, Kyoto University School of Medicine, Shogoin, Sakyo-ku, 606, Japan E-mail: akio{at}kuhp.kyoto-u.ac.jp
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and methodsResultsDiscussionReferences |
|---|
To clarify the functional role of human pre-supplementary motor area (pre-SMA) in `cognitive' motor control as compared with other non-primary motor cortices (SMA-proper and lateral premotor areas) and prefrontal area, we recorded epicortical field potentials by using subdural electrodes in five epileptic patients during presurgical evaluation, whose pre-SMA, SMA-proper, prefrontal and lateral premotor areas were defined by electric cortical stimulation and recent anatomical orientations according to the bicommissural plane and callosal grid system. An S1-Go/NoGo choice and delayed reaction task (S1-choice paradigm) and a warned choice Go/NoGo reaction task (S2-choice paradigm) with inter-stimulus intervals of 2 s were employed. The results showed (i) transient potentials with onset and peak latencies of about 200 and 600 ms, respectively, after S1 in the S1-choice paradigm mainly at pre-SMA and to a lesser degree at the prefrontal and lateral premotor areas, but not in the S2-choice paradigm. At SMA-proper, a similar but much smaller potential was seen after S1 in both S1- and S2-choice paradigms and (ii) slow sustained potentials between S1 and S2 in both S1- and S2-choice paradigms in all of the non-primary motor areas investigated (pre-SMA, SMA-proper and lateral premotor areas) and prefrontal area. It is concluded that pre-SMA plays a more important role in cognitive motor control which involves sensory discrimination and decision making or motor selection for the action after stimuli, whereas SMA-proper is one of the main generators of Bereitschaftspotential preceding self-paced, voluntary movements. In the more general anticipation of and attention to the forthcoming stimuli, non-primary motor cortices including pre-SMA, SMA-proper and lateral premotor area, and the prefrontal area are commonly involved.
pre-SMA; selection; voluntary movements; subdural recording; choice paradigm
BP = Bereitschaftspotential; CNV = contingent negative variation; SEP = somatosensory evoked potential; SMA = supplementary motor area
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Currently the supplementary motor area (SMA) in non-human primates is divided into the rostral SMA (pre-SMA) and the caudal SMA (SMA-proper) on the basis of physiological and anatomical criteria (Wiesendanger, 1986
; Rizollati et al., 1990; Tanji, 1994). It has been reported from studies using the single cell recording technique that in primates pre-SMA neurons discharged more frequently in preparation for the forthcoming reaching movements than SMA-proper neurons, and the latter were more related to execution of the reaching movements (Matsuzaka et al., 1992). Approaches to understanding the function of human pre-SMA and SMA-proper have included functional imaging studies such as PET (as reviewed by Picard and Strick, 1996), and more recently functional MRI (fMRI) (Hikosaka et al., 1996; Humberstone et al., 1997). The results from these studies suggested that a part of the mesial frontal lobe situated rostral to the line drawn through the anterior commissure (AC) perpendicular to the line connecting the AC and the posterior commissure (PC), i.e. the VAC line (Talairach and Tournoux, 1988), was significantly active with respect to the higher order aspects of motor control such as intrinsic movement selection, motor learning, complex movement and Go/NoGo trials, independent of the body part involved (Deiber et al., 1991; Shibasaki et al, 1993; Hikosaka et al., 1996; Picard and Strick, 1996; Humberstone et al., 1997). Thus, this area is regarded as the pre-SMA in humans. These imaging results, however, also contain a variety of factors related to voluntary motor execution such as attention, discrimination, judgement, decision, preparation, anticipation, execution, assessment of performance and somatosensory feedback input mainly due to their limited temporal resolution.
Previously we recorded EEG from the SMA-proper in association with spontaneous, voluntary movements in epilepsy patients by using subdural electrodes (Ikeda et al., 1992, 1993, 1995, 1996a). This technique best delineates the temporal characteristics of the cortical electrical activity in a particular area covered by the subdural electrodes, and the results demonstrated that the SMA-proper generated Bereitschaftspotential (BP) (Kornhuber and Deecke, 1966; Shibasaki et al., 1980) in accordance with its somatotopy. Subdural EEG recordings in epilepsy patients by employing a warned S2-choice reaction paradigm were also made (Ikeda et al., 1996a, b; Hamano et al., 1997). We found that (i) in dichotomous discrimination tasks which required decision making with regard to the S2 stimuli, bilateral mesial frontal cortices rostral to SMA-proper generated transient field potentials and (ii) orbitofrontal and mesial prefrontal cortices generated slow potentials in the period of uncertainty and anticipation preceding the forthcoming informative stimulus (S2). Hamano et al. (1997) further investigated these decision-related potentials and slow sustained potentials in a larger number of patients who had subdural electrode grids at the lateral prefrontal, temporal, parietal and occipital association cortices. These negative transients were not particularly evident at the primary motor or sensory area, but were widely observed at the mesial and basal prefrontal, lateral and mesial parietal, lateral and mesial temporal, and mesial occipital areas, all of which belonged to non-primary motor, non-primary sensory or association cortices. In addition, slow shifts between S1 and S2 were seen at the prefrontal, lateral premotor, SMA, primary sensorimotor, mesial temporal and occipital association areas.
In the three studies described above (Ikeda et al., 1996a, b; Hamano et al., 1997), however, three problems have not been resolved. First, with regard to the mesial frontal area, SMA-proper was defined by electric cortical stimulation, but any further distinction between the SMA-proper, pre-SMA and mesial prefrontal area, especially between the latter two areas, was not clear. In our study (Ikeda et al., 1996a), among the many subdural electrodes analysed on the mesial frontal cortex anterior to the SMA-proper, two were located 20 mm anterior to an electrode which corresponded to the SMA-proper of the face area, and they generated a transient potential after dichotomous S2 stimuli with a peak latency of 350 ms. However, it was unclear as to whether these two particular electrodes belonged to pre-SMA or mesial prefrontal area as anatomical analysis in relation to the VAC line was not performed. A clear distinction among these structures is important when considering the functional role of these individual areas. Secondly, the paradigm adopted in all three studies described was a warned S2-choice Go/NoGo reaction paradigm, which required the subjects to make dichotic discrimination, selection and decision, and also to instigate motor execution immediately upon the S2-Go signal. Therefore, it was not possible to differentiate between the discrimination process and the final decision for motor execution as to Go or NoGo. The distinction would only be possible by adopting the S1-choice paradigm and comparing it with the results of S2-choice paradigm. Thirdly, no studies have been done with regard to subdural potentials at the lateral prefrontal, lateral premotor and primary motor areas in association with an S1-choice reaction-time paradigm. We, therefore, conducted the present study paying special attention to the resolution of these three, especially of the first two, concerns.
Recent analyses of single cell recordings in non-human primates demonstrated that, in contrast to SMA-proper cells, pre-SMA cells generated transient activity when shifting motor plans in response to the instruction signal (Matsuzaka and Tanji, 1996), and a similar specific activity in pre-SMA was seen when updating a motor task (Shima et al., 1996). Updating or shifting a motor task encompasses a selection process which is strongly related to motor preparation. Therefore, in order to clarify how closely the human pre-SMA, strictly defined by the currently accepted anatomical and functional criteria, is associated with cognitive motor control in voluntary movements when compared with SMA-proper, prefrontal and lateral premotor cortices, we analysed epicortical field potentials directly recorded from human cortex in epilepsy patients. Preliminary results have been reported elsewhere in an abstract form (Ikeda et al., 1997). The function of human pre-SMA in generating BP preceding spontaneous, voluntary movements based on findings obtained in patients 1 and 2 of the present series has been described elsewhere (Yazawa et al., 1997b, 1998). Two other patients (patients 3 and 4) were discussed (Mikuni et al., 1997; Mima et al., 1997) and ictal subdural EEG findings in patients 1–3 were reported elsewhere for entirely different purposes (Ikeda et al., 1999).
|
Material and methods |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Subjects
We recorded field potentials directly from the surface of the frontal cortex in five right-handed patients with medically intractable partial seizures (two men and three women, age ranging from 21 to 53 years with the mean of 33 years). All five patients were clinically evaluated for epilepsy surgery by using subdural electrodes, and informed consent was obtained from all patients after the purpose and possible consequences of the studies were explained, which were according to the Clinical Research Protocol No. 79 approved by the ethical committee of Kyoto University School of Medicine.
The cortical electrical potentials were recorded by using chronically implanted subdural electrodes (AD-TECH) made of platinum. Each electrode was 3 mm in diameter, and the centre-to-centre inter-electrode distance was 1 cm. This invasive technique helps to identify (i) the extent of the epileptogenic region by seizure recording, and (ii) the function of the cortex around the epileptogenic region by electrical cortical stimulation and by recording somatosensory evoked potentials (SEPs) (Hahn and Lüders, 1987).
The electrodes were placed at both the SMA-proper and pre-SMA in three patients (patients 1 and 2 on the right hemisphere, and patient 5 on the left), the right mesial prefrontal area in one patient (patient 1), the lateral prefrontal area in two patients (patients 2 and 3, right and left hemisphere, respectively), the left lateral premotor area in one patient (patient 4) and the right primary motor hand area in one patient (patient 2). The anatomical location of these electrodes was defined as described in the Methods section on `Cortical mapping' (see below). Four out of five patients had an increased signal abnormality on T2-weighted MRI which was later confirmed by surgery; in the right cingulate cortex (gliosis) (the area corresponding to electrode 3 and its inferior portion, that is partly extended and close to electrode 10 in Fig. 2) in patient 1, the right lateral premotor cortex (astrocytoma grade II) in patient 2, in the left middle to inferior frontal gyrus (astrocytoma grade II) in patient 4 and in the high lateral convexity of the left frontal lobe (cortical dysplasia) in patient 5. The potentials recorded from the electrodes placed over these lesions were not used for the further analysis.
|
Experimental paradigms
Two reaction-time paradigms using paired auditory stimuli (S1 and S2) were employed in the present study: (i) warned choice Go/NoGo task (S2-choice reaction-time paradigm) and (ii) delayed response paradigm following dichotomous choice Go/NoGo task (S1-choice reaction-time paradigm). In both paradigms voluntary movements were adopted as a response task to S2.
The S2-choice paradigm was adopted from the authors' previous studies as follows (Ikeda et al., 1994, 1996a, b). A pair of tone bursts of different pitch and of a duration of 20 ms were presented for S1 and S2 with an interval of 2 s. S1 was always a tone burst of 1000 Hz, while S2 was either a tone burst of 1500 Hz (Go) or that of 2000 Hz (NoGo). Go and NoGo stimuli were presented in a random order with the same probability, and the motor task employed was middle finger extension of the hand contralateral to the side of the implanted electrodes. The patients were instructed to respond as quickly as possible upon the S2-Go signal, but not to respond upon S2-NoGo. The EEG segment from 1 s before the S1 onset to 1.5 s after the S2 onset was averaged for Go and NoGo trials separately, and the group average across Go and NoGo trials were made. The next warning signal was set to be delivered at variable intervals between 3.5 and 7.5 s after the onset of each S2 signal.
In the S1-choice paradigm, which we adopted partially from another study of ours (Lai et al., 1997), S1 was either a tone burst of 1500 Hz (Go) or of 2000 Hz (NoGo) while S2 was always a tone burst of 1000 Hz. Go and NoGo stimuli were presented in a random order with the same probability. The subjects were asked to react or not to S2 as quickly as possible, by the same motor task as used for the S2-choice paradigm depending on the S1. The EEG segment analysed, the averaging method and the inter-trial interval were the same as those employed for the S2-choice paradigm.
The patient was asked to keep quiet during each recording session and to postpone the next task if he accidentally moved prior to the task movement. Before recording sessions, the patient was given a training period until the examiner was satisfied that the subject consistently produced brisk movements which were preceded and followed by complete muscle relaxation. During the recording, trials in which apparently significant artefacts and erroneous responses by the patients were noted were rejected manually to store by a special on-line rejection paradigm (Ikeda et al., 1994, 1996a, b). One recording session typically lasted 5–6 min, and was repeated four to five times with an intermission of a few minutes between sessions.
Data acquisition and analysis
Cortical recordings were done simultaneously from 20–32 subdural electrodes in each patient. The subdural electrodes were all referenced to a scalp electrode placed on the mastoid process contralateral to the side of implantation. The bandpass filters applied for EEGs and EMGs were 0.016–60 Hz and 20–60 Hz, respectively. The sensitivity of EEG was set to 2 mV full scale (12 bit). Electro-oculograms (EOG) recorded from an electrode placed lateral to the right outer canthus which was referenced to the mastoid electrode was monitored simultaneously with the same filtering and sensitivity as those used for EEG recording.
All of the electrographic output signals were digitized at the sampling rate of 200 Hz per channel, and stored on magneto-optical disks by a compact evoked potential recording machine (DP1100, NEC San-ei) with the aid of a special purpose computer program designed for subsequent off-line analysis. In addition to the special on-line rejection program as described above, an off-line rejection program was also applied; trials associated with either (i) erroneous response by the subjects, or (ii) movements which were not brisk enough to identify a clear EMG onset, were excluded from subsequent analysis. A total of 72–127 trials were selected for each of the Go and NoGo conditions of S1- and S2-choice paradigms separately. For each subject, a group average waveform was obtained for each condition after confirmation that the two ensemble averaged EEGs were reproducible as in Fig. 2C
. Subsequently Go and NoGo trials were averaged altogether for each of the S1- and S2-choice paradigms.
Patients 1, 3, 4 and 5 had EEG recordings for both paradigms, and patient 2 had only the S1-choice paradigm because of the patient's condition.
Cortical mapping
For cortical mapping of the mesial frontal surface, electrical cortical stimulation and the recording of SEPs were done before the present experiment. Each subdural electrode was individually stimulated to identify cortical function (Lüders et al., 1987). Details of the methodology for stimulation and the subsequent cortical mapping have been described elsewhere (Lüders et al., 1987; Ikeda et al., 1992). The SMA-proper was identified by its unique responses on the mesial surface, consisting of predominantly tonic motor responses of the upper as well as lower limb and of the trunk, neck and face, either unilaterally or bilaterally. These motor responses were somatotopically organized within the SMA-proper (Lim et al., 1994). A negative motor response was defined as the cessation of voluntary tonic muscle contraction or rapid alternating movements without loss of awareness during the stimulation (Lüders et al., 1987). Cortical SEPs to electrical stimulation of the median nerve at a rate of 1.1 Hz were recorded from the same subdural electrodes.
In order to define the precise anatomical location of pre-SMA and SMA-proper in patients 1 and 2, a plain lateral view of skull X-ray film, taken after implantation of the electrodes, was first superimposed upon the sagittal view of the T1-weighted MRI taken before surgery by using common landmarks, i.e. nasion and inion, and the sizes adjusted to each other for each patient (Ikeda et al., 1995, 1996a). In patient 5, T1-weighted MRI of the sagittal view was taken when the patient had the subdural electrodes in the skull, to define the precise location directly. Based on the atlas of Talairach and Tournoux (1988) and on the histological analysis in human SMA (Zilles et al., 1996), the electrodes located on the mesial surface of the superior frontal gyrus just anterior to the VAC line were judged to be on the pre-SMA. If the electrodes placed on the mesial surface of the superior frontal gyrus showed a negative motor response by electrical stimulation, then they were also judged to be on the pre-SMA regardless of their relative location to the VAC line. However, the electrodes which elicited a positive motor response when stimulated were excluded from pre-SMA. Based on the atlas of Talairach and Tournoux (1988) and on the recent PET studies (Picard and Strick, 1996), the anterior border of pre-SMA was further defined by the line 20 mm anterior and parallel to the VAC line. In patients 1, 2 and 5 it was also confirmed that the above boundary between the pre-SMA and prefrontal areas was consistent with the callosal grid system (Lehman et al., 1992; Olivier, 1996) in which the boundary was provided by the line (anterior callosal plane) drawn at the anterior edge of the genu of the corpus callosum perpendicular to the plane through the lower border of the genu and splenium of the corpus callosum.
If the electrodes on the mesial surface of the superior frontal gyrus were located posterior to the VAC line, and also if these electrodes did not elicit a negative motor response, then those electrodes were judged to be on the SMA-proper. Since a negative motor response by electrical stimulation is not necessarily elicited in every subject even if the subdural grid electrodes are located on the rostral part of the superior frontal gyrus (Lüders et al., 1987), the presence of a negative motor response was not regarded as essential for defining pre-SMA in the present study.
|
Results |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Cortical mapping
Patient 1 had four 1 x 4 electrode strips at the right mesial frontal area, and another 1 x 4 electrode strip containing electrodes 5–7 was placed at the boundary between the mesial and lateral parts of the superior frontal gyrus (Figs 1A and 2A and B
). Patient 2 had one 2 x 8 electrode grid at the right mesial frontal area, one 4 x 5 grid at the right lateral frontal area and one 4 x 5 grid at the right perirolandic area (Figs 1B and 3). Patient 3 had one 2 x 8 electrode grid at the left lateral prefrontal area (Fig. 4), one 2 x 8 grid at the left basal frontal area, and one 2 x 8 grid at the left mesial to lateral temporal area. Patient 4 had two 4 x 5 electrode grids at the left lateral premotor area (Fig. 5) and left inferior to middle frontal gyrus. Patient 5 had one 2 x 8 electrode grid at the left mesial frontal area and one 4 x 5 grid on the left lateral frontal convexity (Figs 1C and 6).
|
|
|
|
|
On electrical cortical stimulation, electrode 2 in patient 1 induced tonic abduction of the left arm, and electrode 3 elicited a negative motor response in the left arm (Fig. 2A and B
). Other electrodes in patient 1 did not elicit any visible response. In patient 2, electrode 1 elicited tonic muscle contraction of the left arm and foot, electrode 4 elicited negative motor responses involving all extremities, tongue and eye movements, and electrodes 2, 3 and 5 did not elicit any response (Fig. 3). Electrode 8 elicited a clear positive motor response in the left arm. In patient 3, electrodes 5, 9, 10 and 13 did not induce any response after electrical stimulation (Fig. 4). In patient 4, electrodes 2 and 4 elicited negative motor responses in the right arm and tongue, respectively (Fig. 5). Electrode 5 elicited a positive motor response in the tongue. Electrodes 1, 3 and 6 did not cause any response. In patient 5, electrodes 1–3 elicited no responses after cortical stimulation. No stimulation was carried out for electrodes 4–6. Positive motor responses were observed at electrode 7 (tonic contraction of the left lower trunk muscles) and electrodes 8 and 9 (tonic abduction of the bilateral feet, more on the left). Positive sensory symptoms in the whole body were elicited by the stimulation of electrode 10 (Fig. 6). With regard to cortical SEPs, a negative peak (+73 ms, 37 µV) was seen at electrode 1 after the left median nerve stimulation in patient 1. In patient 2, electrodes 1–5 showed positive peaks (either +55 ms or +88 ms, 20–30 µV) after the left median nerve stimulation. In patient 5, a broad, small negative peak (+30 ms, 6 µV) was seen at electrodes 7–9 after the right median nerve stimulation.
From the anatomical point of view, the VAC line in patient 1 was located between electrodes 3 and 4, and electrodes 11–13 were located at or within 20 mm rostral to the VAC line, and therefore were on or caudal to the anterior callosal plane according to the callosal grid system (Fig. 1A). The VAC line in patient 2 was between electrodes 2 and 4 as well as between electrodes 3 and 5 (Fig. 1B). In patient 5, the VAC line was on electrodes 2 and 5. Brain MRI taken with the implanted electrodes clearly showed that electrodes 1–3 were located on the superior frontal gyrus. Electrodes 1 and 4 were located posterior to the anterior callosal plane (Fig. 1C).
Cortical mapping was made by taking all the above findings into account as follows. In patient 1, electrodes 1 and 2 were judged to be on the SMA-proper, electrodes 3–9, 11 and 12 on the pre-SMA, electrodes 14–16 on the prefrontal area and electrodes 10 and 13 on the cingulate sulcus (Figs 1A and 2A and B). Electrode 3 showed continuous slow activity most likely due to the epileptogenicity, and consequently the averaged waveform did not show reproducible responses. In patient 2, electrodes 1–3 were judged to be on the SMA-proper, electrodes 4 and 5 on the pre-SMA, and none were on the cingulate gyrus (Figs 1B and 3). In addition, electrodes 6 and 7 were judged to be on the lateral prefrontal cortex, and electrode 8 on the right primary hand motor area. In patient 3, electrodes 5, 9, 10 and 13 were judged to be on the lateral prefrontal area (Fig. 4). In patient 4, electrodes 2 and 4 were judged to be on the lateral premotor area corresponding to the primary (lateral) negative motor area (Lüders et al., 1995), and electrode 5 on the primary motor face area (Fig. 5). In patient 5, electrodes 1 and 2 were judged to be on the pre-SMA, electrodes 3 and 7–10 on the SMA-proper, electrode 4 on the cingulate sulcus and electrodes 5 and 6 on the cingulate gyrus (Figs 1C and 6).
Behavioural data
As we monitored surface EMG activity instead of button press the onset times of the rectified EMG in the S1- and S2-choice paradigms were 135 ms and 335 ms in patient 1; 130 ms and 175 ms in patient 3; 135 ms and 250 ms in patient 4; 270 ms and 335 ms in patient 5, each respectively, and 180 ms for the S1-choice task in patient 2.
The reaction time in the S1-choice paradigm was much shorter than that in the S2-choice paradigm in all patients except patient 2 who underwent only the S1-choice paradigm. We could not provide the precise error rate in the present study as we rejected clearly erroneous responses during recording from storage by a special on-line program as in the previous studies (Ikeda et al., 1994, 1996a, b). However, the number of erroneous trials in both S1-and S2-choice paradigms was 0, 1 or 2 in a 6 min recording session consisting of 40 trials, and thus an error rate of less than 5% was estimated. We questioned all patients after the experiment about their own strategies to utilize S1 signals in both S1- and S2-choice paradigms, and we confirmed that, in the S2-choice paradigm, all patients paid considerable attention to S1 in order to be able to respond as quickly as possible on S2-Go.
Subdural potentials in the mesial frontal area
Patient 1
In the S1-choice paradigm, large negative or positive transients were seen after S1 mainly at the pre-SMA (electrodes 6, 7, 8 and 12) and transients of smaller amplitude were observed at the prefrontal (electrodes 15, 16) and ill defined at the SMA-proper (electrode 2) (Figs 2A and 7A). These former transients had an onset latency of 205 ms and peaked at 630 ms after the S1 onset with the amplitude of 50 µV at electrode 8 and they were clearly present equally in both S1-Go and S1-NoGo trials (Fig. 2C). The transient potentials seen at the SMA-proper and the prefrontal area had a similar onset latency (190 ms at electrode 2 and 210 ms at electrode 15), but their peak latency was slightly shorter (425 ms and 580 ms) than that at pre-SMA (electrode 8). In contrast, in the S2-choice paradigm, negative transients after S1 were completely absent at both the pre-SMA and the prefrontal area (Figs 2B and 7A). At the SMA-proper, however, negative transients of similar latency and smaller amplitude as in the S1-choice paradigm were seen.
|
Clear negative slow sustained potentials occurring between S1 and S2 were seen at pre-SMA (electrodes 6–9) not only in the S1-choice but also in the S2-choice paradigm (Figs 2A and B and 7B
). These were also equally present in Go and NoGo trials in both S1- and S2-choice paradigms (Fig. 2C). They started at 305 ms and 205 ms after the S1 onset in the S1-choice and the S2-choice paradigm, respectively, reached their peak ~1000 ms after S1 (electrode 9) (Fig. 2A and B) and were sustained until the S2 onset. Similar slow, but smaller potentials were also seen at the prefrontal area (electrode 15) and SMA-proper (electrode 1) equally in both paradigms (Fig. 2A and B).
Only in the S2-choice paradigm in which the dichotic discrimination and the selection were done on S2, were negative transients seen after S2 and those were mainly at the pre-SMA (electrodes 6, 7, 8 and 12) with an onset latency of 200 ms and peak latency of 600 ms after the S2 onset (Fig. 2B). The amplitudes were larger in S2-Go trials than in S2-NoGo trials (Fig. 2C), as the potentials recorded after S2-Go contained activity related to both discrimination/selection and motor execution, but the potentials recorded after S2-NoGo were related only to discrimination/selection, similar to those recorded immediately after S1-Go/NoGo in the S1-choice paradigm.
Patient 2
As only the S1-choice paradigm was employed in patient 2, we could not compare the differences in response after S1 between the two tasks. However, upon both S1-Go and S1-NoGo, in general a large positive activity with an onset and peak latency of 165 ms and 310 ms, respectively, was seen at the pre-SMA (maximum at electrode 4). At the SMA-proper, smaller (electrode 2) or ill defined (electrode 1) positive transients were seen (Figs 3 and 7A). Among the electrodes at the pre-SMA, electrode 4 showed the largest negative sustained slow potential which started about 500 ms after S1. It was clearly seen in S1-Go trials, but was also seen in S1-NoGo trials to a lesser degree (Figs 3 and 7B). SMA-proper (electrode 2) also generated a negative slow potential starting at ~450 ms after S1, but it was smaller than at the pre-SMA (electrode 4) and was seen only during the S1-Go trials.
Patient 5
In the S1-choice paradigm, large positive and negative transients of more than 75 µV were seen at electrodes 1 (pre-SMA) and 4, respectively, after S1, and they peaked at 500 ms after S1. These potentials were seen in both S1-Go and S1-NoGo trials (the waveforms in Fig. 6 are combined S1-Go and S1-NoGo sessions). These results suggest the presence of a dipole generator between electrode 1 (pre-SMA) and 4. These transient potentials are not very obvious at electrodes 3 and 7–10 (SMA-proper) (waveforms at electrodes 7–10 are not shown in Fig. 6), but similar, smaller ones are apparent at electrodes 5 and 6 (anterior cingulate gyrus). In contrast, in the S2-choice paradigm, there were no positive and negative transients after S1 at pre-SMA (electrode 1) which is similar to the situation observed with patient 1 (Fig. 6). At the SMA-proper (electrode 3), and at the anterior cingulate gyrus (electrodes 5 and 6), however, negative transients of small amplitude were seen similar to those in the S1-choice paradigm, and their amplitudes (~50 µV) were as small as ones seen after S1 in the S1-choice paradigm.
Slow sustained potentials occurring between S1 and S2 were ill defined in the S1-choice paradigm at any electrode. In the S2-choice paradigm, small, slow potentials were seen only at electrodes 4–6 which were mainly located in the anterior cingulate gyrus (Fig. 6). The potentials started about 1500 ms after the S1 onset, and gradually increased in amplitude.
In the S2-choice paradigm in which dichotic discrimination and response selection were done upon S2, large positive and negative transients of more than 75 µV were seen after S2 at electrodes 1, 4 and 5 (mainly at pre-SMA). These had a peak latency of 500 ms after S2. These transient potentials after S2 were similar to those after S1 in S1-choice paradigm, in terms of the location and temporal relationship to the stimulus, which were strongly related to the discrimination and selection process to dichotic stimuli (Fig. 6)
Subdural potentials in the lateral frontal area
Patient 2
In the lateral prefrontal area (electrodes 6 and 7), upon S1 in the S1-choice paradigm, small positive-negative complexes were seen with a peak latency of initial positivity at 175 ms after S1. These were followed by clear negative slow shifts starting at 440 ms after S1, and they were sustained until the S2 onset. In the primary hand motor area (electrode 8), there was no clear potential seen between S1 and S2, and a large negative transient was seen after S2 only in the S1-Go trial. The onset and peak latency of this transient were 165 ms and 350 ms after S2, respectively, which were earlier by about 50–150 ms than those of rectified EMG activity (Figs 3 and 7).
Patient 3
In both S1-Go and S1-NoGo trials in the S1-choice paradigm, large negative transients were seen after S1 mainly at electrodes 9 and 5 (Fig. 4). Electrode 9, located more rostrally than electrode 5, showed a similar response to S1 in the S2-choice paradigm, but it was slightly smaller than in the S1-choice paradigm. Electrode 5 showed no such similar potentials after S1 in the S2-choice paradigm. Electrodes 5 and 10 showed sustained slow potentials between S1 and S2 in both S1- and S2-choice paradigms.
Patient 4
Large positive and negative complexes were seen at electrodes 4 and 5, respectively exclusively upon S1 in the S1-choice paradigm and upon S2 in the S2-choice paradigm and the peak latencies were 280 ms and 195 ms after S1 as well as after S2 (Fig. 5). Between S1 and S2, slow positive potential was seen at electrode 1 in the S1-choice paradigm, and at three more electrodes (electrodes 3, 4 and 5) in the S2-choice paradigm (Fig. 5). In both S1- and S2-choice paradigms, electrodes 3 and 6 showed large transients of negative and positive complexes (peak latency at 140 ms and 215 ms, respectively) immediately after both S1 and S2.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Discrimination/selection-related potentials
The present results are summarized in a schematic illustration in Fig. 7
. Transient potentials after S1 (Figs 7A and 8) had an onset latency of ~200 ms, and the peak latency varied from ~300–600 ms, but was mainly ~600 ms. They were often observed as large responses of more than 50 µV in amplitude and were either negative or positive in polarity, when recorded by referential montage. It was clearly demonstrated that, as seen in patients 1, 2 and 5, among mesial frontal cortices, pre-SMA generated the largest transient potentials after S1 in the S1-choice paradigm but not in the S2-choice paradigm. The prefrontal area showed a similar tendency. In contrast, the SMA-proper generated much smaller potentials after S1 in both the S1- and S2-choice paradigms. This suggests that pre-SMA and also the prefrontal cortex are active in the discrimination of dichotic stimuli related to motor action even though the actual execution is carried out 2 s later, whereas SMA-proper is related to simple arousal or attention to the warning stimulus. Therefore, the potentials originating from the SMA-proper most likely correspond to the scalp-recorded early component of contingent negative variation (CNV) observed in a simple warned reaction paradigm (Walter et al., 1964; Tecce and Cattanach, 1993). In patient 2, with whom only the S1-choice paradigm was employed, the transient potential after S1 was seen at pre-SMA (electrode 4 in Fig. 3) but not after S2. Therefore, that particular potential is also most probably related to dichotic discrimination and response selection. It is also worthwhile mentioning that the lateral premotor area, as seen in patient 4, showed a similar tendency, although to a lesser extent than with the pre-SMA.
|
Reaction time in the S1-choice task in patient 1 (135 ms) was shorter than in patient 5 (270 ms), but the transient potential immediately after S1 in the S1-choice task did not show clear differences in the onset- or peak time between the two patients. Reaction time in the S1-choice paradigm reflects several sequential processes including sensory discrimination of the auditory S1 signal, the linkage to a motor task, the selection of motor outcome, the storage of the task and simple motor reaction in response to the S2 cue signal. Therefore, the discrimination/selection-related potentials of the same temporal profile among the patients could result in the different reaction times as with the present patients.
In the S2-choice paradigm in which dichotic discrimination and response selection upon S2 and motor execution upon S2-Go were done, similar transient potentials were seen after S2 (electrodes 6, 7, 8 and 12 in patient 1 in Fig. 2B; electrode 9 in patient 3 in Fig. 4; electrodes 4 and 5 in patient 4 in Fig. 5; electrodes 1 and 4 in patient 5 in Fig. 6). They were seen at the electrodes at which the transient potentials after S1 in the S1-choice paradigm were observed, although the waveforms were distorted probably due to the simultaneous occurrence of the movement-related potentials after S2 as observed in Fig. 2C. This observation also supports the concept that the transient after S1 described above is discrimination- and selection-related.
Recently it was suggested from animal experiments using unit recording (Shima et al., 1991) and by activation studies of humans (Fink et al., 1997) that the cingulate motor area was also involved in voluntary movements. In the present study, only the findings at electrodes 5 and 6 in patient 5 are consistent with this idea. Electrodes 8 and 9 in patient 1, which showed the largest discrimination/selection-related potential and sustained type of response, respectively, were located in the superior frontal gyrus, and in patient 5, positive and negative transient activities were located at electrodes 1 and 4, both of which were on or superior to the cingulate sulcus, and none of the electrodes in patient 2 were on the cingulate gyrus. Therefore, the results obtained in the present study are not representative of the function of the cingulate motor area.
Previously when we compared scalp-recorded CNV between the S1- and S2-choice paradigms, a significantly large, frontally dominant early CNV component was observed in the S1-choice paradigm (Lai et al., 1997). In that study, upon dichotic S1 or S2, the simple Go/NoGo signal was not employed, but either extension or flexion of the right wrist was chosen as a motor task. Therefore, the difference of the scalp distribution between the S1- and S2-choice paradigms was observed only in the time segment corresponding to the early CNV. Scalp electrodes at the frontal midline area would not pick up the potentials arising from the mesial part of the bilateral superior frontal gyri because of the tangential direction of dipoles to the scalp electrodes and the cancellation between the two dipoles of the bilateral opposite polarity (Lang et al., 1991). However, as seen in patient 1, electrode 6 (Figs 1A and 2A) located rostral to the VAC line on the boundary between the mesial and lateral part of the superior frontal gyrus was one of the largest generators of this transient potential among the electrodes in the pre-SMA at the mesial part of the superior frontal gyrus. This suggests that the functionally determined pre-SMA is not restricted to the mesial part of the superior frontal gyrus. It was also shown by Luppino et al. (1991) that cytoarchitectonically and electrophysiologically defined SMA-proper (F3) and pre-SMA (F6) partly extended from the mesial surface to the dorsal surface of the frontal lobe. Therefore, the present results could lead us to a conclusion that the differences of the early CNVs observed, using the scalp electrodes at the frontal midline area, between the S1- and S2-choice paradigms (Lai et al., 1997) was most likely due to the activity of the pre-SMA and prefrontal cortices.
With regard to the specific function of the pre-SMA, recent single cell recording in non-human primates demonstrated that pre-SMA cells, but not SMA-proper cells, generated transient activity when shifting motor plans in response to the instruction signal (Matsuzaka and Tanji, 1996). Similar specific activity in pre-SMA was seen only when updating the motor task between the different series of sequential motor tasks (Shima et al., 1996). In the study by Matsuzaka and Tanji (1996), 1.5–3 s before the impending signal, a 50 Hz tone burst (switching signal) was delivered to monkeys in response to which they had to switch from the current motor task to a new one. Thirty-one per cent of the neurons in the pre-SMA showed a shift-related activity of 500 ms duration after the switching signal, whereas only 7% of the SMA-proper had the same neuronal activity. This result is consistent with the present findings in the pre-SMA and SMA-proper. Since shifting or updating a motor task before the motor execution contains similar features to the S1-choice paradigm employed in the present study in terms of discrimination, selection and decision for motor plan, it is most likely that one of the specific functions of the pre-SMA is `cognitive' processing strongly related to the forthcoming motor tasks, i.e. `cognitive motor control' in voluntary movements. They also showed that phasic responses to visual cue signals which indicate the direction of the forthcoming arm-reaching movements were more abundant in the pre-SMA than in the SMA-proper (Matsuzaka et al., 1992). This may be consistent with the present results in terms of pre-setting the movement direction upon the external signal.
A recent study in humans using fMRI showed that, in the simple Go/NoGo paradigm, the mesial frontal area rostral to the VAC line (pre-SMA) had an increased signal in association with both the Go and NoGo trials, and the area caudal to the VAC line and rostral to paracentral lobule (SMA-proper) was activated only with Go trials (Humberstone et al., 1997). Since that particular study employed a single stimulus paradigm in which discrimination, selection, decision making and then motor execution, if needed, occurred together in response to each stimulus, and since there was only 3 s of time resolution in spite of the single event analysis method, it could not differentiate between stimulus- and motor-related activities. Nevertheless, at least from the view point of Go/NoGo discrimination, it suggests an important role of the pre-SMA as in the present study.
The transient potentials after S1 seen at the mesial prefrontal areas were smaller than those in the pre-SMA, as seen in patient 1. Previously we showed that, in S2-choice paradigms, the similar transient potentials after S2 (decision-related potentials) were observed in the mesial prefrontal area (Ikeda et al., 1996a). Since the pre-SMA received the direct input from the prefrontal cortex (Luppino et al., 1993), it is important to link the information about behavioural content to actual processes for motor execution (Fuster, 1989; Passingham, 1996). Therefore, it is likely that the discrimination and selection processes related to motor execution are conducted in both prefrontal and pre-SMA, and the latter could augment it downstream of the information processing.
Slow shifts between S1 and S2
Slow potentials between S1 and S2 were obvious mainly at the pre-SMA, and also at the prefrontal SMA-proper and lateral premotor areas to a lesser extent. They were seen both in S1-choice and S2-choice paradigms (Fig. 7B). They started occurring immediately after the disappearance of discrimination/selection-related potentials, or gradually became apparent ~400 ms after S1 signal. There were two types of developing pattern: (i) as seen in patients 1 and 2, potentials peaked at ~1000 ms after S1 and were sustained with the same amplitude until the S2 signal, or they slightly diminished in amplitude immediately before the S2 signal; (ii) mainly as seen in patients 3–5, slow shifts appeared at ~1000 ms or even later, and gradually increased in amplitude before the S2 signal. These waveforms are consistent with the late CNV recorded by scalp electrodes (Walters et al., 1964) as well as by subdural electrodes in our previous studies (Ikeda et al., 1996a, b). The waveforms were well documented in S1-NoGo trials in the S1-choice paradigms (as shown at electrode 9 in patient 1 in Fig. 2C and at electrode 4 in patient 2 in Fig. 3), strongly suggesting that these non-primary motor and prefrontal areas are active in very close association with general anticipation or attention for the forthcoming stimuli, but are not strongly or at all related to motor preparation at least between S1 and S2 in this particular paradigm. The S2-choice paradigm causes more uncertainty before S2 than the S1-choice paradigm, and thus the magnitude or the extent of these slow potentials may be proportionally larger in the S2-choice paradigm. In the lateral premotor area in patient 4 (Fig. 5), the S2-paradigm involved three more electrodes generating slower shifts than in the S1-paradigm, but a similar tendency was not seen at the pre-SMA or prefrontal area in patients 1 and 2.
Since the waveforms of these slow potentials are consistent with scalp-recorded late CNV, the present study supports the concept that scalp-recorded late CNV consists of subcomponents, and these are not related just to simple motor preparation in the setting of CNV paradigm. We also previously showed that orbitofrontal, mesial prefrontal, SMA (Ikeda et al., 1996a, b), lateral premotor, primary sensorimotor, mesial temporal and occipital association areas (Hamano et al., 1997) generated slow potentials between the two signals by using the S2-choice reaction-time paradigm. In the present study, we further clarified the involvement of the pre-SMA as one of the most active areas to generate these slow shifts among those areas, and that since it was active in S1-NoGo trials as well as in S2-NoGo trials, it was most likely related to general anticipation, aside from motor preparation. It is worthwhile stressing here that the present results do not deny that pre-SMA is also involved in motor preparation or execution. Recently we demonstrated that pre-SMA generates BPs before spontaneous voluntary movements regardless of the site of the movements (Yazawa et al., 1997b), whereas SMA-proper generates BPs in good accordance to its somatotopy (Ikeda et al., 1992). The primary motor hand area in patient 2 in the present study did not show subdural potentials before S2 in S1-Go trials, but immediately after S2 the clear transient potentials were seen. Therefore, the present paradigms, i.e. CNV paradigms with a paired auditory choice reaction-time setting, would be suitable to elucidate slow potentials unrelated to motor preparation in the analysis segment between S1 and S2.
In preparation for self-paced, voluntary movements, both pre-SMA and SMA-proper generated BPs without clear differences in onset time between each, although the pre-SMA was always active regardless of the site of movements (Yazawa et al., 1997b). However, once `cognitive' processes such as discrimination, decision or response selection elicited by external stimuli were introduced in association with motor control of voluntary movements, the significant functional difference between pre-SMA and SMA-proper was clearly demonstrated. When functionally mapping the cortex in areas other than primary motor and sensory areas by using chronically implanted subdural electrodes, currently used techniques such as electrical cortical stimulation and short-latency evoked potentials to sensory stimulation are by no means sufficient. BP recording could be useful for mapping the primary motor cortex (Yazawa et al., 1997a), but it may not be clinically sufficient for motor mapping of the non-primary motor cortices in the mesial frontal lobe (Allison et al., 1996). When precisely delineating the sophisticated areas of the frontal lobe in relation to higher functions (Goldman-Rakic, 1987; Fuster, 1989) in the field of epilepsy surgery, the analysis of cortical evoked potentials by using choice paradigms as in the present study might be promising for clinical use.
|
Acknowledgments |
|---|
The authors wish to thank Dr Geoff Barrett for his comments on this manuscript. This study was supported by Grants-in-Aid for Scientific Research (A) 09308031, (A) 08558083, on Priority Areas 08279106 and (C) 10670583 from Japan Ministry of Education, Science, Sports and Culture, Research for the Future Program from the Japan Society for the Promotion of Science JSPS-RFTF97L00201, Research Grant for Treatment of Intractable Epilepsy and General Research Grant for Aging and Health from Japan Ministry of Health and Welfare, and Research Grant from Epilepsy Research Foundation.
|
References |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Allison T, McCarthy G, Luby M, Puce A, Spencer DD. Localization of functional regions of human mesial cortex by somatosensory evoked potential recording and by cortical stimulation. Electroencephalogr clin Neurophysiol 1996; 100: 126–40.[Medline]
Deiber M-P, Passingham RE, Colebatch JG, Friston KJ, Nixon PD, Frackowiak RS. Cortical areas and the selection of movement: a study with positron emission tomography. Exp Brain Res 1991; 84: 393–402.[ISI][Medline]
Fink GR, Frackowiak RS, Pietrzyk U, Passingham RE. Multiple nonprimary motor areas in the human cortex. J Neurophysiol 1997; 77: 2164–74.
Fuster JM. The prefrontal cortex. New York: Raven Press; 1989.
Goldman-Rakic PS. Circuitry of primate prefrontal cortex and regulation of behavior by representational memory. In: Mountcastle VB, Plum F, editors. Handbook of physiology; Sect. 1, Vol. 5, Pt. 1. Bethesda: (MD). American Physiological Society; 1987. p. 373–417.
Hahn JF, Lüders H. Placement of subdural grid electrodes at the Cleveland Clinic. In: Engel J Jr, editor. Surgical treatment of the epilepsies. New York: Raven Press; 1987: 621–7.
Hamano T, Lüders HO, Ikeda A, Collura T, Comair YG, Shibasaki H. The cortical generators of the contingent negative variation in humans: a study with subdural electrodes. Electroencephalogr Clin Neurophysiol 1997; 104: 257–68.[Medline]
Hikosaka O, Sakai K, Miyauchi S, Takino R, Sasaki Y, Pütz B. Activation of human presupplementary motor area in learning of sequential procedures: a functional MRI study. J Neurophysiol 1996; 76: 617–21.
Humberstone M, Sawle GV, Clare S, Hykin J, Coxon R, Bowtell R, et al. Functional magnetic resonance imaging of single motor events reveals human presupplementary motor area. Ann Neurol 1997; 42: 632–7.[ISI][Medline]
Ikeda A, Lüders HO, Burgess RC, Shibasaki H. Movement-related potentials recorded from supplementary motor area and primary motor area: role of supplementary motor area in voluntary movements. Brain 1992; 115: 1017–43.
Ikeda A, Lüders HO, Burgess RC, Shibasaki H. Movement-related potentials associated with single and repetitive movements recorded from human supplementary motor area. Electroencephalogr Clin Neurophysiol 1993; 89: 269–77.[ISI][Medline]
Ikeda A, Shibasaki H, Nagamine T, Terada K, Kaji R, Fukuyama H, et al. Dissociation between contingent negative variation and Bereitschaftspotential in a patient with cerebellar efferent lesion. Electroencephalogr Clin Neurophysiol 1994; 90: 359–64.[ISI][Medline]
Ikeda A, Lüders HO, Shibasaki H, Collura TF, Burgess RC, Morris HH 3rd, et al. Movement-related potentials associated with bilateral simultaneous and unilateral movement recorded from human supplementary motor area. Electroencephalogr Clin Neurophysiol 1995; 95: 323–34.[ISI][Medline]
Ikeda A, Lüders HO, Collura TF, Burgess RC, Morris HH, Hamano T, et al. Subdural potentials at orbitofrontal and mesial prefrontal areas accompanying anticipation and decision making in humans: a comparison with Bereitschaftspotential. Electroencephalogr Clin Neurophysiol 1996a; 98: 206–12.[ISI][Medline]
Ikeda A. Lüders HO, Shibasaki H. Generation of contingent negative variation in the supplementary sensorimotor area. Adv Neurol 1996b; 70: 153–9.[ISI][Medline]
Ikeda A, Terada K, Yazawa S, Shibasaki H, Kimura J, Mikuni N, et al. Subdural recording of slow potentials with choice reaction paradigm in human frontal lobe [abstract]. Neurosci Res 1997; Suppl 21: S191.
Ikeda A, Taki W, Kunieda T, Terada K, Mikuni N, Nagamine T, et al. Focal ictal direct current shifts in human epilepsy as studied by subdural and scalp recording. Brain 1999; 122: 827–38.
Kornhuber HH, Deecke L. Hirnpotentialänderungen bei Willkürbewegungen und passiven Bewegungen des Menschen: Bereitschaftspotential und reafferente Potentiale. Pflüger Arch Ges Physiol 1965; 284: 1–17.
Lai C, Ikeda A, Terada K, Nagamine T, Honda M, Xu X, et al. Event-related potentials associated with judgment: comparison of S1- and S2-choice conditions in