INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the medial part of the frontal cortex of primates, at least six motor-representation areas have been defined: the primary motor cortex, supplementary motor area (SMA), presupplementary motor area (pre-SMA), supplementary eye field (SEF), and rostral and caudal cingulate motor areas (Matelli et al. 1991; Picard and Strick 1996; Tanji 1996). Among these, the SMA and pre-SMA, which are located largely on the medial wall, have been extensively studied in connection with limb motor behavior (Brinkman and Porter 1979; Matsuzaka and Tanji 1996; Mushiake et al. 1991; Nakamura et al. 1998; Rizzolatti et al. 1998; Shima and Tanji 2000; Tanji 1994). The supplementary eye field (SEF), which is located at the medial-most part of the dorsal surface of the frontal cortex, has been defined as an oculomotor representation area (Schlag and Schlag-Rey 1987).

It has been established that the SMA and pre-SMA are definable based on the effects of low-current microstimulation and neuronal response properties (Luppino et al. 1991; Matsuzaka et al. 1992; Tanji 1994). The SEF has been defined as an area in which saccades are evoked with low thresholds, and neuronal activity during oculomotor tasks has also been reported in this area (Chen and Wise 1995; Fujii et al. 1995; Huerta and Kaas 1990; Mushiake et al. 1996; Olson and Gettner 1996; Russo and Bruce 1993, 1996; Schlag and Schlag-Ray 1987). However, the extent to which the limb and oculomotor representations are uniquely segregated in the two areas is not known. Neuronal activity in both the SMA and pre-SMA has been related to saccades (Mann et al. 1988; Schall 1991), and neuronal activity in the SEF is influenced by limb movement (Mushiake et al. 1996). Although anatomical connectivity distinguishes each of the three areas (Luppino et al. 1993; Parthasarathy et al. 1992; Shook et al. 1990; Stanton et al. 1993; Wang et al. 2001), it is also known that connections to the SEF are not limited to oculomotor-related areas (Huerta and Kaas 1990).

In view of the need to clarify the relative degree of involvement of these three areas in limb versus eye movements, we examined the neuronal activity that occurred while a monkey performed three trained motor tasks: a saccade-only task, an arm-reach-only task, and a combined saccade-reach task. We found that largely task-selective relationships characterized each of the three areas, although some properties of neuronal activity overlapped.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subject and apparatus

Two male Japanese monkeys (Macaca fuscata) were trained to perform the task in this study. The animals were cared for in accordance with the Guiding Principles for the Care and Use of Laboratory Animals of the National Institutes of Health and the Guideline for Institutional Animal Care and Use published by our institute. To perform the trained motor task, each monkey was seated in a primate chair, with its head restrained, facing a panel. The right arm was used for the task, and the left arm was restrained. The panel contained five buttons equipped with red and green light-emitting diodes (LEDs). A central button was placed directly in front of the animal's face, and the other four buttons were placed 7.5 cm above, below, right, and left of the central button. We placed a hold button in front of the primate chair. We monitored eye position and movement with an infrared corneal reflection monitor system (RMS Hirosaki) with a sampling rate of 250 Hz.

Behavioral tasks

We trained the monkeys to perform a conditional motor task that required capturing a target with a saccadic eye movement (saccade-only task), with a combined eye and arm movement (saccade-reach task), or with an arm movement only (arm-reach-only task), according to visual instructions. The three tasks were intermixed throughout the training and recording sessions. A task began when the monkey pressed the hold button and one of the five LEDs was illuminated as a fixation point. The monkey was then required to fixate on the fixation point for 1.5-2 s (fixation period). In the saccade-only and saccade-reach tasks, a second LED was turned on during the fixation period to serve as a cue for the future target. If the monkey maintained fixation on the original LED for an additional 500-800 ms (delay period) after the cue was presented, the original LED was dimmed. This served as the GO signal. If the second LED was red, the monkey was required to capture the target with a saccade (saccade-only task, Fig. 1A). The monkey was rewarded if the second LED was captured with a saccade within 400 ms of the GO signal. If the second LED was green, the monkey was required to both saccade to the target and reach out and touch the target (saccade-reach task, Fig. 1B). In the saccade-reach task, the time difference between saccade onset and arm movement onset was required to be within 300 ms. In the reach-only task, the color of the target changed from red to green after the monkey had maintained fixation on the target for 1.5-2.0 s. This change in color served as the GO signal for the reach-only task, and the monkey was then required to touch the fixation target within 500 ms (Fig. 1C). If it captured the target successfully, a drop of fruit juice was delivered after a 500-ms delay. The inter-trial interval was fixed at 1.5 s.

View larger version (20K): [in this window] [in a new window]   Fig. 1. Schematic illustration of the behavioral task sequence. The task started when the monkey pressed the hold button using its right arm and fixated its gaze on the red fixation point. The target color indicated the type of forthcoming task; if the target was red, the monkey was required to capture the target by a saccade without releasing the hold button (A, saccade-only task); if the target was green, the monkey had to capture the target by saccade and by reaching with the right arm (B, saccade-reach task). If no target signal was presented and the fixation point turned to green, the monkey had to reach for the fixation target without moving its eyes from the fixation target (C, arm-reach-only task).

Eye position was monitored and judged every 4 ms as being within or outside a fixation window with a radius of 2.5° of visual angle around the center of the fixation target (0.5°). If the eye position left this window during the fixation period, the trial was aborted and treated as an error trial. In the saccade-only and saccade-reach tasks, saccade onset was defined as the time when the eye position left the fixation window. In these tasks, saccade offset was defined as the time when the eye position entered the target window, a radius of 2.5° around the target position. The monkey had to finish the saccade within 400 ms after the GO signal. In the reach-only task, the monkey was not allowed to move its eyes until shortly after reaching for the fixation point with its right arm. If it moved its eyes during this period, the trial was an error trial. In the saccade-only task, the monkey was not allowed to move its arm until shortly after capturing the target with a saccadic eye movement. If it did move its arm during this period, the trial was an error trial.

Surgical and recording methods

After completing the initial behavioral training, aseptic surgery was performed under pentobarbital sodium anesthesia (30 mg/kg im) with atropine sulfate. Antibiotics and analgesia were used to prevent postsurgical infection and pain.

Neural activity was recorded using glass-insulated Elgiloy microelectrodes (0.8-1.5 M at 333 Hz). The electrodes were manipulated with a hydraulic microdrive (Narishige MO-81) and inserted through the dura matter. Neuronal activity was discriminated using a conventional custom-made window discriminator. Discriminated unit activity was stored with a record of behavioral events, eye position read-outs, and electromyographic (EMG) data on a PC, which was also used to control the task parameters.

An intracortical microstimulation technique (ICMS: 330 Hz, 0.2-ms duration, 10-80 µA, 10-50 pulses) was used to locate the three different cortical areas: the SMA, pre-SMA, and SEF (Fig. 2). In the SMA, ICMS evoked body movements at relatively low currents (<40 µA). The arm and orofacial areas were found in the rostral part of the SMA. Arm movements were evoked at relatively higher currents in the pre-SMA than in the SMA (<60 µA). Saccades were elicited with currents <80 µA in the SEF. The direction of the SEF-evoked saccades was contralateral to the stimulated hemisphere. We avoided the use of strong currents for ICMS to avoid both noxious effects and misleading stimulus effects resulting from the spread of electrical current and spurious activation (cf. Tehovnik and Lee 1993).

View larger version (55K): [in this window] [in a new window]   Fig. 2. Spatial location of recording sites in the medial cortex of monkeys A (top) and B (bottom). Far left and far right: a top-down view of the recording sites. The 8 middle rows, rows a-d in the top view, were reconstructed in three-dimensional multiple planes (planes a and b were separated by 1 mm and planes c and d were separated by 0.5 mm). Center: a top-down view of the cortex showing the rostrocaudal and mediolateral extent of the recording sites (rectangles) relative to the cortical sulci. Light blue spindles denote the areas of the cortex where the face is represented [areas where intracortical microstimulation (ICMS) evoked orofacial movements, and neurons responded to somatosensory stimulation of the orofacial region or were active with facial movement]. Red, green, and dark-blue bars denote sites where movement-related neurons were recorded in the supplementary eye field (SEF), supplementary motor area (SMA), and presupplementary motor area (pre-SMA), respectively. Each region was defined according to previously established criteria (Matsuzaka et al. 1991).

Data analysis

Neuronal data were sorted according to the task conditions and displayed on-line as raster plots and peri-event histograms. For the statistical analysis of neuronal activity, we defined four periods according to behavioral events: a control period (lasting 300 ms and beginning 200 ms after fixation onset), a cue period (300 ms after the cue signals), a GO period (150 ms before and after the GO signal), and a movement period (150 ms before and after movement onset). If the neuronal activity during any period was significantly higher than during the control period (Mann-Whitney, P < 0.01), then the cells were classified as task-related cells. In this report, we only describe cells that had significant movement-related activity.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Behavioral analyses

We first compared the mean reaction times (RTs) for saccades and reaches under the three task conditions. The RT for a saccade was defined as the time between the GO signal and saccade onset, and the RT for a reach was defined as the time between the GO signal and releasing the hold button. The mean RT for saccade onset during the saccade-only task was 239 ± 32 (SD) ms. The mean RT for hand release during the reach-only task was 284 ± 23 ms. During the saccade-reach task, the mean RTs for saccade onset and releasing the button were 244 ± 41 and 287 ± 40 ms, respectively, and saccade onset tended to precede button release in this condition. There was no significant difference in RTs for saccade onset between the saccade-only task and the saccade-reach task. There was also no significant difference in RTs for button release between the reach-only task and the saccade-reach task.

Physiological identification of the medial premotor areas

Before analyzing the neuronal activity of the two monkeys while they performed the tasks, we attempted to delineate the SMA, pre-SMA, and SEF on the basis of ICMS effects and on neuronal responses to somatosensory and visual stimulation. We examined neuronal responses to joint manipulation or to muscle tapping and also to cutaneous stimuli (brushing or touching of the glabrous or hairy skin). Figure 2 summarizes the neuronal-activity recording sites assigned to the SMA, pre-SMA, and SEF based on the effects of ICMS and the properties of their neuronal responses to sensory inputs (Fujii et al. 1995; Matsuzaka et al. 1992). In the medial wall, we found orofacial region at the rostral end of SMA and the area further rostral to this region was assigned as the pre-SMA. The pre-SMA neurons responded to visual stimuli more often than the SMA neurons. When ICMS was applied, longer trains of pulses were required to evoke movements in the pre-SMA than the SMA and if movements were evoked, movements were more complex in the pre-SMA than SMA. In the dorsal convexity lateral to the pre-SMA, we found a region where ICMS evoked saccadic eye movements, and this part of the cortex was defined as SEF.

On the medial wall, two forearm regions were found rostrocaudally (the SMA and the pre-SMA). These areas were separated by a small area in which orofacial movements were evoked (Fig. 2). The threshold currents needed to evoke arm movements were higher in the rostral arm region than in the caudal arm region. Neither saccadic nor slow eye movements were evoked in these regions. In the medial dorsal convexity, apart from the medial wall, ICMS did not evoke arm or hand movements, but currents smaller than 80 µA evoked saccades in a limited portion of the cortical surface (SEF in Fig. 2). The saccade-evoking area was 3-7 mm anterior to the orofacial area of the SMA in the medial wall and 4-7 mm from the midline.

Neuronal database

We recorded movement-related activity in 337 SEF neurons, 335 pre-SMA neurons, and 192 SMA neurons. These neurons showed significantly more activity during the movement period than in the control period (Mann-Whitney P < 0.01) under at least one of the three task conditions. The neuronal activity was classified into seven groups based on the activity under the three task conditions, as shown in a Venn diagram (Fig. 3, top). The neurons in groups 1, 5, and 7 were selectively active during the saccade-only, saccade-reach, and reach-only tasks, respectively. The neurons in groups 2, 4, and 6 were active during two of the three tasks, as indicated in the Venn diagram. The neurons in group 3 were active during all three tasks.

View larger version (19K): [in this window] [in a new window]   Fig. 3. Classification of movement-related neurons in the SEF, pre-SMA, and SMA. We classified movement-related neurons into 7 categories based on their neuronal activities during the performance of saccade-only, saccade-reach, and arm-reach-only tasks. Each category is shown in the Venn diagram. The bar graph shows the population of each group for the total number of movement-related neurons in each area for the 2 monkeys (A and B). The total number of movement-related neurons is displayed in the top right corner of each panel.

SMA neurons

In the SMA, 127 of the 192 cells (66%) showed changes in activity during the movement period in both the reach-only and saccade-reach tasks. An example of this group is illustrated in Fig. 4A. In addition, a population histogram showing activity of neurons of this type is illustrated in Fig. 6A. The neurons in this group were found predominantly in the left hemisphere (83%), contralateral to the task-performing arm. This arm-reach-selective activity was not influenced by the presence or absence of accompanying saccades (113/127, P > 0.05, Mann-Whitney test). Saccade-related activity was only found infrequently in the SMA, and saccade-related neurons were found near the rostral border (n = 11). In both monkeys, the neuronal activity was predominantly classified as group 6, exhibiting activity changes nonselectively during both the reach-only and saccade-reach tasks (Fig. 3, right).

View larger version (30K): [in this window] [in a new window]   Fig. 4. Examples of the neuronal activity of 1 SMA neuron and 2 pre-SMA neurons. A: an SMA neuron that showed significantly increased firing during tasks that required arm movement while the monkey was performing the arm-reach-only and saccade-reach tasks but not during the saccade-only task. This example is from group 6 in our classification. B: a pre-SMA neuron that showed premovement activity during the arm-reach-only and saccade-reach tasks. This neuron is an example from group 6. C: a pre-SMA neuron that showed a nonselective increase in activity during all tasks (group 3). The raster and histogram are aligned with movement onset (saccade onset was used for the saccade-only task and hold-button-release onset was used for the saccade-reach and arm-reach-only tasks).

Pre-SMA neurons

In the pre-SMA, 335 neurons showed changes in activity during movement. We observed two major types of movement-related activity (groups 3 and 6 in Fig. 3, middle, and Table 1). In the first, observed in 130 neurons (39%), the movement-period activity increased during both the reach-only and saccade-reach tasks. An example of this type of pre-SMA neuron is illustrated in Fig. 4B. The second type of neuron, observed in 119 of the 335 pre-SMA (36%) neurons, showed movement-related activity in all task conditions (Fig. 4C). Population histograms showing neuronal activity in these two groups are illustrated in Fig. 6, B and C. There were only a small number of saccade-related neurons, which were active either during the saccade-only task (n = 27) or during both the saccade-only and saccade-reach tasks (n = 32).

SEF neurons

We recorded 337 cells from the SEF. They are classified in Table 1 (and Fig. 3, left) according to their relationships to the three tasks. About one-third of the SEF neurons (109/337) showed increased activity when the monkeys performed saccades under the saccade-only or saccade-reach conditions. An example of typical saccade-related activity for this type of SEF neuron is shown in Fig. 5A. This type of neuronal activity was considered eye-movement-selective because the activity increased when saccades were performed, and it was not affected by arm movements. However, some saccade-related activities differed according to whether arm movement accompanied eye movement (groups 1 and 5 in Fig. 3 and Table 1). An example of this type of activity is shown in Fig. 5B. The SEF neuron displayed in Fig. 5B was active in the saccade-only task but not in the saccade-reach task (group 1 in Fig. 3 and Table 1). Sixty-one of 337 cells showed this type of selectivity. As observed in the pre-SMA, 91 of 337 SEF neurons exhibited movement-related activity under all task conditions (Fig. 5C). Population histograms of three representative groups of neurons in the SEF are illustrated in Fig. 6, D-F. We also found 46 SEF neurons that were active during the arm-reach-only and saccade-reach tasks but not during the saccade-only task (group 6 in Fig. 3 and Table 1).

View larger version (33K): [in this window] [in a new window]   Fig. 5. Examples of the activity of 3 SEF neurons under the 3 different task conditions. A: an SEF neuron that showed an increase in activity during the saccade-only and saccade-reach tasks but not during the arm-reach-only task (group 2 in Table 1 and Fig. 3). B: an SEF neuron that showed presaccadic activity exclusively during the saccade-only task (group 1 in Table 1 and Fig. 3). C: an SEF neuron that showed premovement activity nonselectively during all tasks (group 3 in Table 2 and Fig. 3). The format of this figure is the same as that of Fig. 4.

View larger version (38K): [in this window] [in a new window]   Fig. 6. Population histograms for 6 subgroups of neurons in the 3 areas shown in Fig. 4 and 5. A: a population histogram for SMA neurons in group 6 (n = 104). B and C: population histograms for pre-SMA neurons in groups 6 (n = 117) and 3 (n = 94). D-F: population histograms for SEF neurons in groups 2 (n = 79), 1 (n = 77), and 3 (n = 72). To construct population histograms, firing rates of individual neurons were normalized for each of 6 groups.

Distribution of saccade and arm-movement-related cells

To display the relative density of the neurons specifically related to saccades and arm-reach movements, we constructed distribution maps of the neurons belonging to groups 2 and 6. As described in the preceding text, neurons in group 2 were active with saccades, regardless of the presence or absence of accompanying arm movements. The neurons in group 6, on the other hand, were active with arm-reach movements, regardless of whether the movements were executed alone or in combination with a saccade. Thus they appeared to be effector-selective and not influenced by behavioral context. The density maps of saccade-related and arm-reach-related neurons are shown in the top and bottom panels of Fig. 7, where the medial wall and the adjacent dorsal convexity of the frontal cortex are depicted as unfolded.

View larger version (71K): [in this window] [in a new window]   Fig. 7. Density plot of eye movement (group 2)- and arm-movement-related neurons (group 6) on the unfolded medial frontal cortex. Top: density map of eye-movement-related neurons. Bottom: density map of hand-movement-related neurons. Left and right: the left and right hemispheres, respectively. To calculate the density of neurons, we combined the data from the 2 monkeys. One pixel represents a 0.5 × 0.5-mm area. The color of each pixel represents different levels of density, as shown in the bottom right corner of each panel. The anterior-posteror (A-P) line of the rostral end of the orofacial region of the SMA is drawn as a horizontal white line in each panel. The 6 points (a-f) in the density map correspond to those in the coronal section of the medial frontal cortex in the right panel. The eye-movement-related neurons were found mainly on the dorsal surface in both hemispheres; arm-movement-related neurons were found mainly on the medial wall. Arm-movement-related neurons were distributed bilaterally in the rostral part, mainly on the contralateral side, in the left hemisphere, on the caudal part of the medial wall.

The density map of saccade cells revealed two major foci, one in the left hemisphere and the other in the right hemisphere. In both hemispheres, the two foci were located on the dorsal surface and they corresponded to the SEFs defined in Fig. 2. Each of the foci was located 5-6 mm anterior to the anterior-posterior (A-P) line of the orofacial region of the SMA, and 3-5 mm lateral to the midline. An additional and less obvious concentration occurred in the left hemisphere, near the rostral part of the left SMA, which corresponds to the orofacial area of the SMA. No focal area for saccade-related neurons was observed in the right hemisphere. In contrast to saccade-related cells, those for arm movement were recorded mainly in the medial wall. We found a bilateral distribution of arm-movement-related cells in the rostral part of the medial wall, which corresponds to the pre-SMA. In contrast, the arm-related cells found in the SMA were mainly in the left SMA, contralateral to the arm used in the task.

Laterality of distribution of arm-related neurons

Figure 7 shows that the laterality of the distribution of arm-related neurons appeared to differ among the three areas. About 83%, or 105 of 127, arm-movement-related SMA neurons were found in the contralateral hemisphere. In contrast, in the pre-SMA, 86/130 (66.2%) arm-movement-related neurons were found in the left hemisphere and 44/130 (33.8%) were found in the right hemisphere. In the SEF, more arm-movement-related neurons were recorded in the left hemisphere (29/46, 63.0%) than the right (17/46, 37.0%). To investigate the laterality of distribution of arm-related neurons more quantitatively, we introduced a laterality index, defined as (N_c  N_i)/(N_c + N_i). N_c indicates the number of arm-related neurons in the contralateral side, and N_i indicates the number of arm-related neurons in the ipsilateral side. Figure 8 illustrates the laterality index for each area. The figure demonstrates that the laterality of the SMA was distinct from that of the pre-SMA and SEF. This difference between the SMA and the other two areas in the distribution of arm neurons in each hemisphere was statistically significant (P < 0.05), but the difference between the pre-SMA and SEF was not.

View larger version (28K): [in this window] [in a new window]   Fig. 8. The laterality index of arm-movement-related neurons (group 6 in Fig. 2 and Table 1) recorded from the SEF, pre-SMA, and SMA. The laterality index is defined as (Nc  Ni)/(Nc + Ni). Nc and Ni are the numbers of arm-movement-related neurons recorded from the left (contralateral) and right hemisphere (ipsilateral), respectively. The laterality of distribution of arm-movement-related neurons was significantly higher in the SMA than in the SEF and pre-SMA (P < 0.01). However, the difference between the SEF and pre-SMA was not significant.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We analyzed neuronal activity in the three medial motor areas during the performance of a behavioral task that required execution of a saccade, an arm-reaching movement, or both a saccade and an arm-reaching movement. We found that neurons in the SMA, pre-SMA, and SEF exhibited area-specific relationships to the performance of the three different tasks. Saccade-related neurons were found predominantly in the SEF, whereas arm-reaching-related neurons were found mainly in the SMA and pre-SMA. On the other hand, we found a group of neurons in the pre-SMA and SEF that were active nonpreferentially in the saccade-only, arm-reach-only, and saccade-reach tasks. We further found that some of the movement-related activity was context dependent, especially in the SEF.

Spatial distribution of effector-selective neuronal activity

We classified neuronal activity into seven categories based on its selectivity for the three tasks. Neurons classified as group 2 were active during the saccade-only and saccade-reach tasks, indicating that their relationship to eye movement was not influenced by the presence or absence of arm movement. Neurons classified as group 6 were active during arm-reach-only and saccade-reach tasks, suggesting their participation in arm movements. We interpreted this to mean that neurons in groups 2 and 6 exhibited effector-selective activity, that is, eye- and arm-selective activity. These neurons were primarily in the SEF and SMA/pre-SMA, respectively, as defined in this study. This agrees with the traditional view that the SEF is primarily an oculomotor area (Russo and Bruce 1996; Schall 1991; Schlag and Schlag-Rey 1987) and that the SMA and pre-SMA are limb-motor areas (Luppino et al. 1991, 1993; Matsuzaka et al. 1992; Piccard and Strick 1996; Tanji 1996). However, small number of neurons were exceptions (5.7% of SMA and 9.6% of pre-SMA neurons categorized as group 2 and 13.6% of SEF neurons categorized as group 6). These events may represent information coming from other areas conveying, for instance, limb-motor information to the primarily oculomotor-related SEF (Mushiake et al. 1996).

The extent to which limb-motor-related neuronal activity is lateralized to the hemisphere contralateral to the task-performing limb is an issue that has not been addressed in behavioral studies in which the SMA and pre-SMA were separately defined (cf. Brinkman and Porter 1979; Tanji and Kurata 1981). For the arm-reach-only task, we found that effector-selective activity was present in 83% of SMA neurons and 66% of pre-SMA neurons in the contralateral hemisphere, indicating that the activity of SMA neurons is more lateralized to a statistically significant degree. However, it should be noted that the laterality is influenced by the behavioral context of the behavioral task (Tanji et al. 1988; also see following text).

Effector nonselectivity and context-dependent activity

Neurons belonging to group 3 were active during all three tasks and therefore interpreted as effector-nonselective. They were more abundant in the pre-SMA (36%) and SEF (27%) than in the SMA (9%). The distribution of these effector-nonselective neurons in the three areas was significantly different (P < 0.05, ² test). The majority of the effector-independent premovement activity did not show visual responses to the appearance of visual targets. Thus the effector-independent activity has little in common with the reactivation responses in V4 (Fischer and Boch 1982). They may be related to target-oriented activity in general or to motor-equivalent activity that has the goal of capturing an object (Mann et al. 1988; Rizzolatti et al. 1988) regardless of the effectors used to capture the object (eyes or arms). More work is necessary to test the validity of this view. The presence of these effector-nonselective neurons, however, suggests a need for cautious interpretation of the neuronal activity in behavioral tasks: the neurons active before initiation of saccades may not always be neurons specifically active with oculomotor tasks: they may actually be effector-nonselective neurons.

On the other hand, the properties of neurons classified as groups 1, 4, 5, and 7 differed from those of groups 2, 3, and 6. Their activity appeared to be conditional, depending on whether the eye or arm movements were performed in isolation or together. Such dependency of neuronal activity on behavioral context was found in our previous report on the SEF (Mushiake et al. 1996). In the present study, context dependency was also found in both the pre-SMA and SMA. Thus in these three areas, some parts of either saccade or arm-movement representations are contingent on whether other body parts are activated together. These findings are reminiscent of a previous work in which digit-movement-related activity in the SMA appeared to be conditional, depending on whether a motor task with one hand was performed in isolation or was accompanied by a task performed with the other hand (Tanji et al. 1987, 1988). Our findings are also relevant to more recent findings that spatially selective neuronal activity in the SEF is dependent on the object-centered location of a cued point, regardless of the absolute location of the object on the screen (Olson and Gettner 1996; Olson and Tremblay 2000). Olson's results could be interpreted as indicating that SEF neurons were sensitive to the context in which saccades were performed. In this sense, our data were consistent with the context-dependency of SEF neurons.

Cognitive aspects of movement-related activity

Studies of the anatomical connectivity of the three areas have revealed differences that suggest roles for the pre-SMA (Bates and Goldman-Rakic 1993; Luppino et al. 1993; Matelli and Luppino 1996) and SEF (Huerta and Kaas 1990; Parthasarathy et al. 1992) in more cognitive aspects of behavior. The behavioral aspects in which the SEF has been implicated include sequential saccades (Gaymard et al. 1993), conditional oculomotor association (Chen and Wise 1995), antisaccade performance (Schlag and Schlag-Rey 1987), reward prediction and detection (Amador et al. 2000), and performance monitoring (Stuphorn et al. 2000). On the other hand, it has been proposed that the pre-SMA has roles in changing planned motor acts (Matsuzaka and Tanji 1996), learning sequential visuospatial procedures (Nakamura 1998), and planning and regulating multiple movements (Shima et al. 1996, 2000). Our findings on context-dependency and effector nonselectivity suggest an additional aspect of behavioral contingency, to which future studies on the functional roles of these medial motor areas should be directed.