JOURNALOF NEUROPHYSIOLOGY Vol. 66, No. 3, September 199 1. Printed in U.S.A.
Neuronal Activity in the Primate Premotor, Supplementary, and Precentral Motor Cortex During Visually Guided and Internally Determined Sequential Movements HAJIME
Department
SUMMARY
MUSHIAKE, MASAHIKO INASE, AND JUN TANJI of Physiology, Tohoku University, School of Medicine, Aoba-Ku, Sendai, 980, Japan
AND
CONCLUSIONS
1. Single-cell activity was recorded from three different motor areas in the cerebral cortex: the primary motor cortex (MI), supplementary motor area (SMA), and premotor cortex (PM). 2. Three monkeys (Macaca .fuscata) were trained to perform a sequential motor task in two different conditions. In one condition (visually triggered task, VT), they reached to and touched three pads placed in a front panel by following lights illuminated individually from behind the pads. In the other condition (internally guided task, IT), they had to remember a predetermined sequence and press the three pads without visual guidance. In a transitional phase between the two conditions, the animals learned to memorize the correct sequence. Auditory instruction signals (tones of different frequencies) told the animal which mode it was in. After the instruction signals, the animals waited for a visual signal that triggered the first movement. 3. Neuronal activity was analyzed during three defined periods: delay period, premovement period, and movement period. Statistical comparisons were made to detect differences between the two behavioral modes with respect to the activity in each period. 4. Most, if not all, of MI neurons exhibited similar activity during the delay, premovement, and movement periods, regardless of whether the sequential motor task was visually guided or internally determined. 5. More than one-half of the SMA neurons were preferentially or exclusively active in relation to IT during both the premovement (55%) and movement (65%) periods. In contrast, PM neurons were more active (55% and 64% during the premovement and movement periods) in VT. 6. During the instructed-delay period, a majority of SMA neurons exhibited preferential or exclusive relation to IT whereas the activity in PM neurons was observed equally in different modes. 7. Two types of neurons exhibiting properties of special interest were observed. Sequence-specz$icneurons (active in a particular sequence only) were more common in SMA, whereas transitionspeczficneurons (active only at the transitional phase) were more common in PM. 8. Although a strict functional dichotomy is not acceptable, these observations support a hypothesis that the SMA is more related to IT, whereas PM is more involved in VT. 9. Some indications pointing to a functional subdivision of PM are obtained. INTRODUCTION
The primate neocortex possesses a remarkable degree of regional specialization. In the visual cortex, for instance, a number of subfields have been defined on the basis of neuroanatomic or physiological criteria. Each subfield is charac0022-3077/9
1 $1 SO Copyright
0
terized by properties of neuronal activity in response to specific aspects of visual input. Motor areas have also been divided into an increasing number of subfields (Barbas and Pandya 1987; Dum and Strick 199 1; Matelli et al. 1985; Rizzolatti et al. 1988; Wiesendanger 198 1; Wise 1987). Next to the primary motor cortex (MI), two major fields in the nonprimary motor cortex have been the subject of extensive studies: the premotor cortex (PM) in the lateral hemispheric surface and the supplementary motor area (SMA) in the mesial cortex. Although both the PM and SMA send projections to MI, these two areas receive largely separate inputs from the thalamus (Matelli et al. 1989; Schell and Strick 1984; Wiesendanger and Wiesendanger 1985) and from other cortical areas (Arikuni et al. 1988; Barbas and Pandya 1987; Jurgens 1984; Matelli et al. 1986). It is, therefore, not surprising that the effects of localized lesions of these areas in experimental animals (Brinkman 1984; Halsband and Passingham 1982; Passingham et al. 1989) and in clinical cases of human subjects (Freund and Hummelsheim 1985; Watson et al. 1986) appear different. Studies utilizing measurements of regional blood flow (Ingvar and Philipson 1977; Roland et al. 1980) and the studies on cerebral motor potentials (Deecke and Kornhuber 1978; Libet et al. 1983) also suggest differences in the way these two areas are involved in motor performance. However, it has not been established whether any differences exist in the activity properties of cells in the PM and SMA. Four recent studies attempted to compare neuronal activity in these two areas (Kurata and Wise 1988; Okano and Tanji 1987; Romo and Schultz 1987; Thaler et al. 1988), but all of these reports revealed similarities rather than differences of activity properties in simple motor tasks. Neurons in both PM and SMA are involved in execution and preparation of forelimb movements triggered by sensory signals or initiated by the self. These studies pointed to a need for examining neuronal activity in more complexly organized motor performance. Of particular interest is the involvement of the nonprimary motor areas in sequential motor tasks. Previous reports (Laplane et al. 1977; Luria 1966) and more recent studies (Brinkman 1984; Marsden 1987) suggest a different manner in which lesions of the SMA and PM affect performance of sequential movements. Moreover, recent DC potential studies suggest prominent involvement of the SMA in sequential or rhythmic movements (Lang et al. 1988, 1990). On the other hand, recent lesion studies on monkeys (Thaler and Passingham
199 1 The
American
Physiological
Society
705
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (192.236.036.029) on September 4, 2018. Copyright © 1991 American Physiological Society. All rights reserved.
706
MUSHIAKE
1989) demonstrated that the presence or absence of visual guidance was the crucial factor in revealing defective motor performance after SMA or PM lesion. Thus we attempted to compare neuronal activity in the two areas in association with a sequential motor task in two conditions: visually guided task (VT) or internally determined task (IT). The precentral motor cortex was also surveyed in the same animal. A considerable degree of regional differences will be described in this report. A preliminary account of this study has appeared previously (Mushiake et al. 1990).
ET
AL.
animal, there were no central LEDs such asthose that triggered the internally guided sequence in the other two monkeys. Instead, all four LEDs in the push buttons were illuminated, which signaled the animal to start the sequential key-pressmovement. The predetermined sequence included top-bottom-right, top-left-right, and top-right-bottom. All of these sequences of motor tasks were controlled by a microcomputer.
Data recording and processing
After the monkey had achieved a consistent correct performance rate of >95% in the task, a stainless steel recording chamber was attached to the skull under aseptic conditions. DurMETHODS ing the surgical operation the monkey was anesthetized with ketamine hydrochloride ( 10 mg/kg im) and pentobarbital sodium (30 Behavioral procedures mg/kg im). Intramuscular injections of antibiotics and pentazoThree male Japanese monkeys (Macacajkcata) were used in tine were given to prevent infection and to relieve postoperative this experiment. Each was seated in a primate chair and was discomfort. Daily care and treatment conformed to NIH guidetrained to place its right hand sequentially on three out of four lines (1985). After the monkey had recovered fully, a glass-insutouch pads that were attached to a panel and placed at a distance lated Elgiloy alloy microelectrode was inserted into the target zone of 30 cm in front of the animal. Two of the four touch pads were through the dura mater with an electronic stepping microdrive positioned vertically and the remaining two were positioned hori- (MO-95 1, Narishige) for extracellular recording. The same eleczontally across a small, centrally placed light-emitting diode trode was used for intracortical microstimulation (ICMS). Con(LED). An additional touch pad was placed at a distal end of an ventional chronic single-unit recording methods were employed. armrest. Each trial was initiated when the animal pressedthis pad For the ICMS, a train of 12 cathodal pulses of 0.2-ms duration at (called a hold key in this paper) with the right hand for a period of 333 Hz was applied at an intensity of ‘.. .- _- _ _ --.< ____ . ---. _ -_. -
‘..
1,:4..
E
FIG. 2. Movement-related activity in 12 different muscles performed in visually guided (Visual) and internally guided (Internal) motor task. EMG, recorded by wire electrodes, is rectified, digitized, and displayed as dot rasters. Each line of the raster denotes individual trials. Results obtained in 15 trials are summed in the histogram below each raster. Activity of all muscles is aligned at the time of occurrence of 1st key touch (4). FDP, flexor digitorum profundus; EDC, extensor digitorum communis: ECU, extensor carpi ulnar-is; FCR, flexor carpi radialis; BIC, biceps; TRI, triceps; NECK, sternomastoideus; TRAP, trapezius; DELT, deltoid; SSP, supraspinatus; ILIO, iliopsoas; VERT, paravertebral.
digit and hand muscles was relatively late in appearance. In all forelimb muscles recorded, magnitudes of activity changes, especially in the period preceding the first key press, were not different regardless of whether the sequential movement was visually or internally guided. Changes of activity in neck and body muscles were modest and started later than those of forelimb muscles. No muscle activity changes were detectable during the instruction period.
Cortical recording sites Neuronal samples were obtained from three different fields in the frontal agranular cortex of three monkeys, contralateral to the task-performing limb. The three portions corresponded to the MI, PM, and SMA (Alexander and Crutcher 1990a; Dum and Strick 199 1; Kurata and Tanji 1986; Tanji and Kurata 1982). Points of microelectrode entry, viewed from the cortical surface, are indicated in Fig.
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (192.236.036.029) on September 4, 2018. Copyright © 1991 American Physiological Society. All rights reserved.
c7 MONKEY
SEQUENTIAL
MOVEMENTS
AND CORTICAL
MOTOR
709
AREAS
3
ML
ARC
5
12
4 5
3
2 4 2
12 2 6 4 6 3 7 7 16 36
MONKEY
2
MONKEY 3
8116
3
1 MI
6
1
2 3 33 3 9 21 2 2 2 1 1
1
4
4
1 1
4 1
1
ROSTRAL
LATERAL
FIG. 3. Surface maps of the left frontal cortex indicating the location of recording sites and task-related neurons in the 3 monkeys. Total number of neurons related to the task in the penetration is indicated. ARC, arcuate sulcus; CENT, central sulcus; ML, hemispheric midline.
MOTOR CORTEX
unit 1st
Key
unit
1’ Touch
1st
Key
2 Touch
Visual
Internal
400ms FIG. 4. Two examples of MI neurons exhibiting similar activity in association with visually guided and internally guided motor task. Neuronal activity is displayed in 2 formats in this figure and in Figs. 7-9. Each row in the raster display indicates a trial, with dots representing individual single-cell discharges. Each histogram is made by summation of data in each raster.
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (192.236.036.029) on September 4, 2018. Copyright © 1991 American Physiological Society. All rights reserved.
710
MLJSHIAKE
ET AL.
M I
SMA
PM
100
158
60
66
Premovement
46
4g
Period 29 16 10 2
0
_a111
1234567
I
0
6
1234567
56 69
Movement
56 40
Period
34 39
36
26 21
25 9
1111
1234567
l4
III
12
Ill,
1234567
0
1
6
FIG. 5. Distribution of neurons in 3 different motor areas, classified according to their relative relations to visually guided (VT) vs. internally guided (IT) motor task, during premovement (top) and movement (bottom) periods. Classification criteria: 1, exclusive relation to VT; 2, VT B IT (P < 0.00 1); 3, VT > IT (0.00 1 < P < 0.05); 4, VT = IT; 5, VT < IT; 6, VT 4 IT; 7, exclusive relation to IT. Number of neurons is indicated on top of each bar. Ordinate indicates percentage in populations recorded in each area.
3. In the MI, the ICMS evoked apparent shoulder, arm, or hand movements with currents of ~15 PA. In the PM, the ICMS effects were not obvious and often ineffective with currents of ~40 PA. In two monkeys, the PM was surveyed at a site both caudal to the arcuate sulcus and inferior to the postarcuate spur (area F4 and F5 by Rizzolatti et al. 1988 and arcuate premotor area (APA) by Dum and Strick 199 1) and, more medially, at a site lateral to the precentral sulcus (F2 by Rizzolatti). In the third monkey, only the APA was surveyed. In the mesial frontal cortex, penetration sites corresponded to the forelimb part of the SMA (Macpherson et al. 1982; Mitz and Wise 1987; Tanji and Kurata 1982). ICMS effects (predominantly proximal forelimb movements plus occasional hand movements) were observed at some sites but not at others. Rostrolateral to the portion of the SMA where neuronal recordings were made, eye movements were evoked by ICMS. At such penetration sites (presumed to be the supplementary eye field; Huerta and Kaas 1990; Schlag and Schlag-Rey 1985), neuronal sampling was not performed.
Data base for neuronal study In the three monkeys, 338, 438, and 328, task-related neurons were recorded in the MI, PM, and SMA, respectively. The primary aim of this study was to compare the neuronal activity during VT and IT. For this purpose, the activity during the three periods (delay, premovement, and movement period, as defined in METHODS) was quantitatively measured and then compared. As a result of the comparison (statistical analysis is also described in METHODS), neuronal activity was classified into seven groups: 1) exclusively related to VT; 2) markedly more related to VT and only slightly related to IT; 3) relatively more related to VT; 4) equally related; 5) more related to IT; 6) markedly more related to IT; and 7) exclusively related to IT. No neurons sampled in MI, PM, and SMA for the present study were related to saccadic eye movements. Because the data obtained in the third animal were not different from those in the other animals, all data were combined and described as a whole.
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (192.236.036.029) on September 4, 2018. Copyright © 1991 American Physiological Society. All rights reserved.
SEQUENTIAL
MOVEMENTS
AND
PM
CORTICAL
SMA
MOTOR
711
AREAS
100
27 24 19
10 6 1 rl 3
Ill
I
I 4
01.
0
0 m,--
0
0
0
12 34 5 6 7 12 34 5 6 7 12 34 6. Distribution of neurons in 3 different motor areas, classified according to their relative relations to the delay-period activity in relation to visually vs. internally guided motor task. Display formats are the same as in Fig. 5.
5
0
0
6
7
FIG.
PM
Unit 1st . .w,*.*..-*.... .
,
,
. .
. V
i
sual
’
.
.
,
w
.
.
..*w.
.
.
.
.
*
.
..
.
. .
. .
‘..W
.
. . .
.
.
,... .
..
.
. ..I
. . . .
.
.
.,..,.....
.
*.
. .
. .
s
.
.
.
,(**. . .
.
.
,
.
.
.
..-,I)
. . *a
.
. l
. l
.I
. *
.* . . , ,
l
. n* -
.
. mnssv
,,..,,
.
.
.
.**.*~*~+~*~~r)..~.*~)
.
.*(.
.
. ..
.
e,-
. m*
.
. l
.. ..
.
l ..-.m.)...... ., . . --4-c. . * . . . . r.*-,rrr..,* . . . . . -m-,.4. ,.w--*l r-rr~+ .I..... . .
*
-w*-
-o--r. (....
+r..rr.-m.H..-
mm.*
‘.‘..‘.‘.‘
.
.
, .
. . ..I. .. . . .
.
.,
,
.
.
..(.
I....
.* .
. .
..I.
.
.
..,.I.,, . . .
-,m . . .
. .
. . . -.
. .
.
Key
2 Touch
*....
.-
.
. ,.. .
.
.
.
. .
.
.
. .
. . .4 . . . .. * . .e . * . . . . . . . . . * . . . I ,. . . . . . . . . . d *
. .
.
* .
. .
.
. . . . . *.
l .
. l
. . . . . . 0..
.*.,.-@“”
* .
,
. ..I..
. 4 . ..a*... . + . . . . . .a,. . . .I . . . . . *.*. . . .* . . . . . .v . . ,* . . a... . 111.. . z.m.m . ..* . . . . . I. .L_ -6. . . . a .‘,. . I. .*.e.rr.... *... . * . , ,*,,*..*. .,.m**... . . ..“. . . .m.. ,-a., m.. . . . . . . . ..I. .l,”
. -m.
. . . .a
m.4
•.-..r*....‘I)~-,,.*...
m.m..m
1st
,,..-r.,rr(r---.r,......r ,,*r.-#~.m
..,,,,...
, ,,,I..~.., ,.,., m ,,.,,.,, . . ..I . . -...a . . . . I. . ,.. . . . .. . . . . . . . ..m... ... . .. . . . .. , . ,... ,... .,a..... . . .* . . . . ,. , . , . . .M.“...W,W . . . . . “,.*--r..-.rr**. . . . . . . -*.-*I -..a., .rr*. .I . a. * .a ., w., . ,,,. . ,
l ,.
.
..*...*
*,*w .. ... . . .*. *..* , . . *-...m 6 . . .
1”. ,,
.
..**...
I..
, ,, . . . ,.. .,... I * ,..... . .
Unit
1 Touch
Key
.
.
* ,
I .
.
.
.
, .
. .
.
.
.
< . .
.
I
8
I
I
I
I
I
6
I I
:
:
I
I
I I
I
i
I . .
8 I
*
.I.
l
4
: I
.
4 .
I
I
.
. .
I
I
.
.
400ms FIG.
7. Two examples of PM neurons exhibiting preferential (left) or exclusive (right) relation to visually guided motor
task.
Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (192.236.036.029) on September 4, 2018. Copyright © 1991 American Physiological Society. All rights reserved.
MUSHIAKE
712
1st
ET AL.
Unit 1 Key Touch
Unit 2 Key Touch
1st
Visual
..a .a .*
.
*a
I. .
.
.
.
. . .... .. *
400ms FIG.
8. Two examples of SMA neurons exhibiting preferential or exclusive relation to internally guided motor task.
Neuronal activity in MI Figure 4 shows typical examples of activity in MI neurons recorded from sites where the low-current ICMS evoked shoulder (unit 1, left) and hand (unit 2) movements. Unit I exhibited continuous activity increase during the premovement and movement periods. Unit 2 exhibited three peaks of activity increase in association with the sequential movement. In either case, magnitudes of activity changes during the premovement period were not significantly different, whether the sequential movement was visually guided (top) or internally guided (bottom). The magnitudes of activity changes during the movement period were not different, either. A vast majority of MI neurons exhibited responses similar to those exemplified in Fig. 4 and categorized as group 4 responses as a result of the statistical study. As indicated in histograms in Fig. 5, 95% of activity changes in the premovement period (top, right) and 93% of those in the movement period (bottom, right) belonged to group 4. No MI neurons showed exclusive relation to either visually or internally guided movement. Thus it can safely be concluded that the neuronal activity in MI, as a whole, was virtually similar, regardless of whether the sequential movement was visually guided or internally guided. Only a small number of MI neurons exhibited activity changes during the delay period. In 46 out of the 338 task-related MI neurons, the activity during the delay period exhibited either increase or decrease, compared with that in the preceding hold period. The detection of these activity changes was made when the activity in a binwidth
of 40 ms during the hold period deviated from the mean value (in at least 2 consecutive bins) during the hold period (only the l-s period preceding the delay period was used for this calculation) by >2 SD from the mean. Most of the activity changes in this period were observed equally after the two instruction signals that told the animal whether it was in the visual or internal mode (Fig. 6, right).
2. ClassiJication of cells according to relation to visually and internally sequenced movement TABLE
Premovement Period
v >1 v =I
PM
SMA
MI
134 (55.3)
50 (23.1)
46 (19.0)
v