Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by WA STATE UNIV LIBRARIES on 11/20/14 For personal use only.

Activity of presentsal neurons during torque-triggered hand movements in awake primates Y. C . WBNG,W.C. KWAN,AND y. T. ~ ~ U R P M Y Received July 24, 1978 WON(;, Y. C., KWAN,H. Cs, alld MCJRPHY, J. T. 1979- Activity of precentral neurons during torque-triggered hand m o v e m e ~ ~in t s awake primates. Can. J . Physiol. Pharmacol. 57, 174-184. In monkeys performing a handle-repositioning task involving primarily wrist flexionextsnsion (F-E) movements after a torque perturbation was delivered to the handle, single units were recorded extracellularly in the contralateral precentral cortex. Precentral neurons were identified by passive somatosensory stimulation, a~ndwere classified into five somatotopically organized ~sopulations.Based on electromyographic recordings. it was observed that flexors and extensors about the wrist joint were specifically involved in the repositioning of the handle, while many other mr~scleswhich act at the wrist and other forelimb joints were involved in the task in a supportive role. Hn precentral cortex, all neharonal responses observed were temporally correlated to both the sensory stimuli amcl the motor responses. Viwnl stimuli, presented simultaneously ~ ~ i torque t k perturbations, did not affect the early portion of cortical responses to such torque perturbations. In each of the five somatotopically organized neuronal populations, task-related neurons as well as task-unrelated ones were observed. A significantly larger proportion of wrist (F-E) neurons was related to the task. as compared with tlne other, nonwrist (F-E) pop~alations. The above findings were discussed in the context of a hypothesis for the function of precentral cortcx during voltantary limb movement in :wake primates. T h i ~hypothesis incorporates a relationship between activities of populations of precentral neurons, defined with respect to their responses to peripheral events at or about single joints, and movements about the same joint.

Introduction The use of HCMS (Asanurzla and Sakata 1967) concurrently with extracellular unit recordings has resulted i~m 21 spatial description of an input-output relationship in the precentral cortex sf awake monkeys. Initially it was observed in single loci that a honaonymous relationship existed between sensory input5 and motor outputs (Rosen and Asanurna 1972). Thus a discrete cortical region, within which applicatio~mof ICMS produced movement about a distal forelimb joint, receivcd major sensory inputs frorm that joint and its related cutaneous fields. In a series of recent studies (Kwan et al. 1978b; Murphy et al. 1978; IVong et al, 1978) this finding was confirmed and extended to i~scludeall cortical areas representing the entire forelimb. Furthermore, multiplc clusters, each of which consisted of several adjacent vertically arranged loci o%similar input-outpaat properties, wcrc observed in spatially separated regioims. Conjugation of all clusters functionally related to a single forelimb joint produced a ring- or crescent-shaped zone. Finally, all forelimb joint

zones were nested in such a way that contiguous joints were succe(;sively representcd, with proximal joint zones encircling distal orzes. Based on tlac above observations, one may hypotlacsize that thcse somatotopically organized neanronal populations in preceratral cortex are preferentially utilized in voluntary motor tasks which involve the homonynaous joint. In order to develop an appropriate and testable ared diction of this hypothesis, two lactsrs must be taken into csr~sideration.Firstly, it is evident that all voluntary movenaents involve neuronal control of multiple joints. if only for purposes of joint fixation and posture. Secondly, it is known that precentral neurons may exert control over spatially separate pools of motoneurons (Shinoda et a%. 1976), which may control different joints. However, it is also apparent that sucla divergent iratluences of a precentral clustcr are functionally weighted, in a manner which permits HCMS at r i specific site to induce movement about only a single joint (Kwan et al. 1978b). A prediction of the above Imypothesis which takes these factors into account is that the activity of claastcrs of precentral neurons, defined by t $ ~ s r a ~ . , v r s r ~ICMS, c ~ ~ s : intracortical rni~so~tinaulation; passive sensory examination as wrist (F-E) neurons, F-E, flexion-extension; DP, delay period; EMG, electromyograph(ic) ; PSTH, peristirnulus o r periresponse time his- wilt be more closely correlated with a voluntary movement involving hand F-E about the wrist joint tograms~;CWS, central nervous system. 0008-4212/79/020174-11$01.8Oi0 @ 1979 National Research Council of Canada/Conseil natioannl de I-echerches du Canada

175

WBNG ET AL.

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by WA STATE UNIV LIBRARIES on 11/20/14 For personal use only.

tanan will other clusters of neurons in the forelimb control region of precentral cortex. The experiment reported in this paper was designed to study this prediction. We used the technique developed by Evarts (Jaspcr 1958 ; Evarts 1964) of extracellularly recording the activity of single neurons in awake behaving ~nonkeysduring the perforlnance of a stimulus-response task. The task involved repositioning a handle by means of flexioin or extension about the wrist joint immediately after a torque perturbation was delivered to the handle. The analysis specifically examined the relative proportions of task-related neurons in each of the soinatotopically organized neuronal populations in precentral cortex.

Methods

Apparatus and T~ask Monkeys (~Wcrcacaarctoides), sitting in a primate chair with head fixation as shown in Fig. 1, were trained to perform a stimulus-response reaction task for 2-3 h a day. In front of the right arm was an encased computer-colatrolled torque motor unit (Aeroflex). Attached to the bottom end of the shaft of this motor unit was an adjustable manipulandum handle. This motor shaft was aligned aboye the axis of the wrist joint, in saach a way that rotation of the shaft was in the same or opposite direction to wrist extension o r flexion. There were no mechanical stops within the

FIG.1. Drawing s f a monkey (Macaca arctoides) working in the test apparatus. R, a head-restraining device; H. a hydraulic microdrive unit with a microelectrode attached to its tip and the whole unit is mounted over the recording chamber-; J, a juice-delivery tube; TM, a torque motor unit; C , a cast to restrain excessive forearm movements; VM, a videomonitor display unit with the target line and the square cursor shown on its screen. (Drawing courtesy Mrs. E. Rodger. 1

possible range of free wrist rotation. A cast, attached t o the motor unit and positioned under the monkey's forearm, was used to restrain excessive forelimb movements other than wrist extension and flexion. In tinaes of somatosensory stimulation and ICMS, both the cast and the motor unit were removed to allow access to the monkey's forelimb. During a trial of the task, a white vertical target line (0.25 ern wide, 25 cm long) ,and a white square caarsor ( 2 cin side) were shown on a dark background sf a videomonitor screen, which was 100 cm in front of the monkey. The position of the target line was under computer contrc~l,while that of the square cursor was coupled to the rnanipulandlum handle. A general description of the task, including a monkey's correct response in a single trial, is schematically outlined in Fig. 2. A single trial was divided into three time-ordered periods: an intertrial period of variable duration, a preliminary period, and a following data-collection period of 5 s duration. During the intertrial period, the square cursor and the target line were not shown on the videornonitor screen. Their appearance marked the onset of the preliminary period. In this period, the monkey was trained to align the square cursor onto the target line, which is in a position marked 'C' in the top trace of Fig. 2, within a specified DP of 1 duration. This position corresponded to the most natural position of the wrist joint at resting state. After the monkey held the handle in this position for a minimum duration of another second, data cojlection began. One second later, a step torque perturbation of 900 g cm was delivered t o the handle, marked as TORQUE ON' in Fig. 2, in any one of the two directions: extension o r flexion. By means of wrist movement opposing that of the torque perturbation, the monkey restored the handle, and hence the cursor, back to the original position. At the end s f each successful trial, the monkey was rewarded with juice.

-

Experimental Procedures and @ku.ssificatiom of h7curores The present study was performed over a period of 18 months cowc~arrentlywith previoa~sstudies in which a detailed descriptiola of experimental procedures (surgical prepztration, somatosensory stima~lation,PCMS, Bmistolsgical identification) is given (Kwan et al. 197Xb; Murphy et aB. 1978; Wsng et al. 1978). Monkeys achieved an $ 5 + % successful performance after a training period c~f1-2 months. Thereafter, a surgical operation to irllplant a recording chamber and a head restraining device was performed. After a recovery period of about 2 weeks following surgery, recording started without further training. After each precentral neuron recorded was studied in relation to the task, thc unit was classified by its responses to somatosensory stimulation into sets identified by single joints. Because of the nature of the task, which primarily involved wrist flexion and extension movements, the neurons were subclassified into five populations: shoulder, elbow, wrist (F-El, wrist (other), and fingers. Wrist CP-E) neurons were exclusively related to the wrist joint in the extension o r in the flexion direction. Wrkt (other) neurons were wrist rleumns other than wrist (F-E) neurons, for example, neurons responding to ulnar deviation of the wrist. Similarly, shoulder, elbow, and finger neurons were neurons related to the corresponding limb parts. Recordhg T~.clrniqi~c.s Glass-insulated platinum-iridium microelectrodes ( 1 MQ at 1 KHz) were used to record extracellularly and to rnicrostimulate cortical tissues. The action potentials were m-

CAN. 9, PHYSIOL. PHARMACOL. VOL. 57, 1979

TARGET POSITION

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by WA STATE UNIV LIBRARIES on 11/20/14 For personal use only.

I T O R Q U E ON

+

1 kg-cm[

v.

lkg-cm[

,

MOTOR TORQUE

H OARNQDULEE T

Extension

151

HANDLE POSITION

lsO

Flexion a

IQO%[

-a/

+- P R E L I M I N A R Y PERIOD

8

4

DATA

COLLECTION PERIOD

HANDLE VELOCITY

4

FIG. 2. Schenlatic representation of the task and motor performance in a successful trial. Trial begins when the target line and the square cursor appear on the videomonitor screen. R (right), C (center), and 1, (left) represent three different positions of the target line, and correspond to the extended, the natural, and the flexed positions about the wrist joint, respectively. DP, allowable response delay period of 1 s duration. plifiecl (Dagan 2400) and converted into a train of equivalent pulses (maximum timing error less than 1 ms) by a multilevel window discriminator. The output of the discriminator together with all movement parameters, sanlpled at 10-ms intervals, such as torque cleveloped at the motor shaft. angular position, velocity, and acceleration of the handle were digitized, displayed, and stored on tape. Other information, such as a cell's spatial location. its responses to passive somatosensory stinlulation were entered separately on a data sheet, and filed accordingly with the tape entry for further analyses. Intramuscular multiunit EMG records of forelimb muscles including flexor digitoruna superficialis, extensor digitoram communis, flexor carpi radialis, flexor palmaris longus, extensor carpi ulnaris, biceps, triceps, deltoideus, and infraspinatus, were made on selected occasions to document typical responses. Bipolar pairs of insulated nichrome wires (25 pm diameter) with l-mm exposed tips were used in these EMG recordings.

Data A nnlysis A statistical routine (hlurphy et ill. 1974) was adopted to study the average behaviour of cellular responses to repeated stimuli and to detect a significant response after the torque perturbation. Each neuron was tested with repeated torque perturbations in both directions, 10 or nlore times in each direction, in n random sequence. In the first part of the analysis. data obtained in trials with torclues delivered in a single direction were compiled together. Then a PSTJ-I was constructed by aligning these data with respect to either sensory stimuli (torque onset) or motor response ( V opoint in Fig. 2 ) , respectively. This V,,point was selected as a time marker of rnotor response because it ccdnsistently occurred at about 50-70 rns after the onset of the voluntary component of agonist EMG activity (Fig. 3 ) and thus was the first detectable kinematic record of the handle being MOTOR TORQUE

Extent of Cortical Exploratiorl The precentral forelimb areas in two monkeys were ~ ~ r a i formly explored (Wong et al. 1977; Kwan et al. 1978a). Firstly, recordings were made in electrode penetrations which were systematically spaced within 10-mrn square grids. Secondly. cortical area5 adjacent to the forelimb representation, such as mouth, neck, and trunk regions on the lateral, anterior, and medial side, respectively, were also explored. Finally, as the posterior bou~ldaryof the forelimb area was found close to the bottom of the central sulcus, recording tracks of 8-10 mm long sampling at less than 250-pm regular intervals were made in the anterior bank of FIG. 3. Intramuscular EMG sampled from flexor palthe central suIcus. Neurons in cortical region caudal to the precentral forelimb area were also recorcled in these deep naaris longus before and after the delivery of a step torque electrode tracks. The above procedures ascertained that the in the extension direction. An arrow in the EMG trace inentire forelimb area was studied. with the exception of a dicates the onset of the voluntary component of the .above few electrode penetrations that went through blood vessels agonist muscle. On top of the EMG recording is a corresponding kinematic record of the handle position. which prevented successful recordings.

177

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by WA STATE UNIV LIBRARIES on 11/20/14 For personal use only.

W O N 6 ET AL.

wrist antagonist activities. Hence, these rnusc~daractivities could be considered to be related to the alignment of the handle in correcting the overshoot, instead of in responding to the torque perturbation. Kinematic analysis of handie positions revealed a return of the handle at a point marked 'Vo' in the handle position trace of Fig. 2. This V , point corresponded to the rnaxirnum deflection of the handle from the original position. The ~rnaxirnurndeflection of the handle was found to be about 15 k 2.5' (mean -t- standard deviation) in terms of wrist rotating angles, and the V o point occul-red at a mean latency of 140 -t- 30 ms after the torque onset irrespective of the direction sf the torque perturbation. Within this stimulus-responise period, the angular velocity of the handle reached a maximum value of less than 100"j s (Fig. 2 ) . After the T',, point. an additional period of 160 -t- 30 ms was required to complete the return of the handle back to the original Results position in cases of no overshooting. This was folEMG Activities and Kinen-tatic Rvcords lowed by a steady holding of the handle on the tarShortly after the oilset of tlae torque perturbation, get position. These observations were conlsistent with EMG recordings in selected forelimb muscles the results obtained in the EMG recordings of wrist showed active participation of wrist flexors or ex- agonist muscles (Fig. 3). tensors as well as many other forelimb muscles in the rep~sitioningof the handle. These muscles in- Sensory-~ZloforProperty of Cortical Re,~ponses Based on the premise that precentral cortex transcluded those acting about all forelimb joints, includforms inputs into motor patterns, it is appropriate to ing the shoulder. Some muscles were active and others silent. A repeatable pattern of excitation and consider all cortical responses with onset latency less inhibition in the wrist agonist and antagonist mus- or only slightly greater than the voluntary reaction cles, respectively, was observed. The excitation pat- time as having possible causal relationships with tern of the wrist agonist muscles, contraction of both sensory stimuli and motor responses (as defined which would consequently reposition the handle in Methods). Figure 4 illustrates this principle in one against the torque, was composed of a phasic burst of the many typical cortical responses. In Fig. 4A: of activities as early as 30-60 ms and a second one all trials are centered about the onset of sensory at or greater than 70 ms after the onset of the torque stimuli, while in Fig. 4B,the same trials are aligned perturbation (Fig. 3 ) . In a small number of trials, with respect to the occurrence of motor responses less than 10% of the total, an additional burst of a ( V , , point). In each of both PSTMs, a phasic brirst smaller amplitude was observed at an earlier latency, of activities fallowed by a tonic increase of cortical about 15-20 ms after the torque onset. Ira all trials, discharges was observed. Based on the background these bursts of EMG activities in the wrist agonist activities in the prestimulus period (Fig. 4 A ) , the muscles were followed by a tonic excitatory activity mean, m, and the 2 standard deviations above the which lasted until the torclue was released at a later mean, m -1- 2 sd, firing level were calculated and time. O n the other hand, the wrist antagonist mus- are indicated by a broken line and a dotted line, cles were silenced as long as the torque was on and respectively. In both BSTHs, the modification of 2 sd only showed one or two phasic bursts shortly after cortical discharges clearly exceeded the m the torque was released. In a few instances, phasic firing level, which was defined as the level for a sigexcitatory activities in these antagonist muscles were nificant response. It can be observed that the P S W also observed after the onset of the torque, but at a in Fig. 4 A has a higher amplitude and faster rise-time long latency of over 200 ms. These muscular activ- peak than the one ita Fig. 4B. This indicates that for ities were found only in trials in which the monkey, this particular cell. the response is less variably corin an effort to reposition the handle rapidly, overshot related to the sensory stimulation than to the motor the target position. In these trials, one also observed response. In other cells, the converse was observed that the occurrence of the overshoot preceded the to be true.

actively returned by the monkey. Hence, a reaction time was defined as the latency between the onset of torque perturbation and the occurrence of t',, point in the handle position trace. A modification of cortical discharges in the PSTIf was considered a significant excitatory response if the following criterion was satisfied. The criterion was defined as an occurrence of two o r more consecutive 10-ms bins, each of which contained a number of spikes exceeding two standard deviations above the mean number of spikes per bin in the control period prior to the stimuli. In cases of inhibition, two standard deviations below the mean background activities or zero firing level was used instead. If the modification of cortical discharges passed the above test. the cell was classified as a responsive or task-related neuron. In a few special cases, the above-mentioned criterion was found inapplicable. For exa.mple, it was observed that two precentral neuron? increased their discharges after the stimuli in singlets and doublets only. In the PSTHs so formed. the responses clustered together to form a large peak within a single 10-rns bin. These two cells were also classified as task-related neurons.

-+

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by WA STATE UNIV LIBRARIES on 11/20/14 For personal use only.

CAN. J. PHYSIBL. PHARMACBL. VOL. 57, 1979

+ I'

I

+ I ,111 +D II 8 11 +I8 1 1 1 1 lSl I +Ill +

*

I1

+

II II I

1

I 1

118

+IS I 1

I

I. I

1

11

6

8[

I I IIBIB I l l I I I i Ill Ill1 I II I 1 B B I i l l 1 II IlSl II

I

+ I I I1 I 1 1 + U 1 1 1 11 l 1 1 6 +I1I Ill 1 1 11 1 1 + I I S 8 1 L IIIIIIi I S B 8 , II I l + 11 818 I + B I1 1 I I +

1

B I

I

B 1

FIG.4. Analysis of responses of a cortical cell to torque perturbations. Raw data from many trials of the same task condition are displayed in rasters, with trials aligned with respect to (A) sensory stimuli? S, and (8)motor responses, Rf. The motor response is defined kinematically by V o(Fig. 2), which occiirs about 70-90 ms after the onset of the voluntary component of EMG activity in the agonist nluscle (Fig. 3). Each stroke in the rasters represents an action potential. A cross within each trial indicates the occurrence of the motor response (A) and of the sensory stinlulus (R). PSTH of 10-111s bins is derived from the raster below. In the PSTHs, a broken line and a dotted line are drawn to indicate the mean background firing level and 2 standard deviations above the mean firing level, respectively. Each raster and its corresponding PSTM show neuronal activities 500 ms before and after the desiglmated event.

Eflect of Visual Information on Torque-triggered Cortical Responses Under normal conditions, labelled as 'somatic visual' condition in Fig. 5, the handle of the motor unit was coupled t s the square cursor on the videomonitor screen. When the torque was delivered, the handle was displaced and so was the square cursor. The displacement sf the square cursor constituted a visual stimulus. Hence, a test was necessary to study the effect of such visual information upon cortical responses to the torque perturbation. In this test, the square cursor was decoupled from the handle at the moment when the torque was delivered. Thus, the square cursor remained still as long as the torque was on. Trials under this condition, designated as 'somatic only9 condition in Fig. 5, were presented to the monkey in a random sequence among trials under the normal condition. Kinematic analysis of handle position indicated that the timing of the V o point

+

had similar values under both conditions. In addition, EMG recordings of wrist agonist muscles showed unchanged patterns of activities. These latter observations refer t s the early restoring phase but not the late holding phase of the motor responses after the torque perturbation was delivered. Thus, one could compare the early portions of torquetriggered cortical responses obtained under both conditions. For all significant cortical responses, the early portions, that is those with a latency less than 100 ms, were found to be similar under both conditions. Two representative examples, one excitatory and the other inhibitory, are shown in Fig. 5 8 and B, respectively. 'In Fig. 5A, the prestimulus firing levels of both PSTHs are relatively constant. Shortly after the torque onset, an excitatory burst of activities with an onset latency of about PO ms was observed under each of the two conditions. The onset latency

WQNG ET AL.

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by WA STATE UNIV LIBRARIES on 11/20/14 For personal use only.

SOMATIC 9 VISUAL

SOMATIC

I

+ VISUAL

+

16 I 1 I I 1 1 1 1 1 I I II ISHBlllli l Ill l I I I I I I I I I I % I L 111111u111111111111 1 I I i 8 I LIB 1411M111 1111 1111111llI 8 I I I I I I I I I I I I IIIllMlllllM I I I I I I I II 1 l I l l l l 1 l l I l l l l l % 11 111111111 1 1 1 1 1 1 I 1 1 l l 111 l l l B llllllllliID l 1 1 1 1 1 1 1 1 1 1 I 1 1 I 8SlUllllll1lSl 1 1 1 1 I 1 I 1 i l l 11111111111ll II I I I I I I I L I II I II \ I I I I I ! % I1IIIIM1 % II I I I l l i l l 1111 1 1 1 1 I I I 11111111 1 1 1 I I

I I I I I I I I 1 I I i I I I I I Ill I I1 I II11111I 11 111111 Ill111 I III I I I l l l l l l l l 1 11111 IIII I I I I I 1 1 I I 1 1 1 1 It 1 111 1 111111 1 1 1 1 1 1 1 1 1 1 111111111 1 1 1 1 1 1 I 1 II 1 1 1 I 11 I 1 1 1 I l l ! I I l l 1 1 I II IIII I

1

L

I

1

9

9

EXTENSION

FLEXION TORQUE SOMATIC

ONLY

SOMATIC

I

TORQUE

ONLY

4

0

4

IF

FLEXION TORQUE

EXTENSION TORQUE

FIG.5. Effect of visual information upon two representative torque-triggered cortical responses, one excitatory (A) and the other inhibitory 03).In the 'somatic visual' condition, the rnanipulandurn handle is coupled to the square coupling is disconnected at the onset of the torque perturbation. Arrow cursor, while in the 'somatic only'c~ndition~this under each raster marks the onset of torque perturbation. See Fig. 4 for detailed notations.

+

is defined by the time of occurrence of the first bin of a significant response. Both cortical responses in Fig. 5A have similar rise times and amplitudes. Similar results were found in the inhibitory cortical responses illustrated in Fig. 5B. In these cases, the cortical discharges were found to reach a zero firing level at and after 60 ms. All these finding indicated that visual stimuli, presented at the same time as the torque perturbation, did not significantly affect cortical responses due to the latter. This result agrees with the observation that visual inputs to precentral cortex have a longer latency than somatic inputs (Evarts 1966, 1973) ,and with EMG observations concerning long-loop reflexes in man (O'Riain 1978).

Proportion of Task-reiated Nerrrons among Different Sonaatotopicakky Brguizized Populations In the two monkeys, a total of 779 precentral neurons were studied in detail. Of this total population,

691 neurons were identified by passive sensory exan~inatioilas having a consistent relationship with contralateral forelimb parts. The remaining 88 cells were found by similar means to be located in cortical areas adjacent to the forelimb representation, such as hindlimb, trunk, neck, and jaw areas. When the activities of this latter group of cells were examined under the task condition, it was found that none of them exhibited a modificatioil of cortical discharges consistently related to either the torque perturbations or the subsequent motor activities. The spatial locations sf task-related and task-unrelated units for one of the monkeys are shown in Fig. 6. A similar spatial pattern was observed in the second monkey. Of the 691 forelimb neurons, 55 units (8.0%) were found to be related to two contiguous joints. namely 20 shoulder-elbow neurons, 16 elbow-wrist neurons, and 19 wrist-finger neurons. Within these classes of neurons, three ( 1595 ) , four (25 9%9 , and

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by WA STATE UNIV LIBRARIES on 11/20/14 For personal use only.

180

CAN. J. PHYSIBL. PHARMACOL. VOL. 57, 1979

figures were derived from findings of previous studies (Wong et al. 1878; Kwan et al. 1978b; hlurphy et aH. 1978). Combining these values with the total number of neurons studied in each population (column 3 ) one could calculate the density of cells studied in each cortical zone (column 6). Values between 2.1 8 and 2.70/nnm+ere obtained. This observation suggests that each cortical zone was sampled and studied evenly. Each somatotopically organized population contained both task-related neurons and task-unrelated ones. The result is listed in columns 1 and 2. Shoulder, elbow, wrist (other), and finger neurons are collectively termed nonwrist (F-E) neurons, as opposed to the remaining group labelled as wrist (F-E) neurons. Out of 580 nonwrist (F-E) neurons, a total of 186 task-related units were recorded. These 186 task-related neurons were made up of 40 shoulder, 60 elbow, 38 wrist (other), and 48 finger neurons. The task-unrelatcd ones included 153 shoulder, 113 elbow, 57 wrist (other), and 71 finger neurons. Of the wrist (F-E) population, 39 cells were related to the task and the other 17 were not. These findings clearly indicated two points: that not every wrist (F-E) neuron was involved in the task utilizing wrist F-E movements, and that not every neuron which is involved in a task which involves primarily wrist F-E moveinents is a wrist (F-E) neuron. Frc. 6. (A) Top view of the left hemisphere of one TO find out whether the wrist (F-E) and nonmonkey showing the cortical areas (bounded by broken wrist (F-E) populations behaved differently in this lines) investigated. ( B ) A parasagittal section (taken along prinlarily wrist-oriented task, relative proportions 1-1 in A ) showing locations of task-related and task-unrelated cells. Each dot along the reconstructed tracks repre- of task-relatcd neurons in all five somatotopically sents a site of recording. Unit responses to natural somato- organized populations were calculated and are listed sensory stimulation are indicated on the left and unit re- in the fourth columll of Table 1. It was found that sponses to the torque perturbation on the right. NR, no the percentagc values varied from population to popresponse; S, shoulder; E, elbow; W,, wrist (flexion); We, ulation in a descending order from wrist (F-E), to wrist (extension); W,, wrist (other); I;, fingers; 11, task unfingers, wrist (other), elbow, and finally to shoulder related; fe, excitation after F-E torques, respectively (inhibition indicated by underlined symbols). (C) Locations of population. Furthermore, a statistical analysis based all tracks plotted on the cortical surface. Large dots indicate on chi-square test was used to examine the distributracks containing task-related neurons and small dots indicate tion of task-related and task-unrelated cells in each tracks where no task-related cell was found along the entire population. The analysis indicated that a significantly penetration. larger proportion sf wrist (F-E) than nonwrist (F-E) neurons was related to the task (x2 = 30.8; eight(40c%) task-related units were found, respec- p < 0.005). The analysis also showed that of all tively. These neurons were not included in Table 1, populations, shoulder neurons were least involved in which neurons having a relationship about single in the task (x2= 26.0; p < 0.005). joints are listed. Of the 636 single-joint neurons, 225 (35.6% ) showed significant modifications of Discussion discharge after the torque was delivered. They are The objective of this study was to investigate the listed in Table I according to their somatotopic identities. In addition, the areal sizes (in square mil- proportion of task-related units in each of the five lirnetres) of cortical zones representing these pop- somatotopically organized neuronal populations in ulations are also given (solunln 5 ) . These latter the precelatral cortex of awake monkeys.

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by WA STATE UNIV LIBRARIES on 11/20/14 For personal use only.

WOWG ET AL.

TABLE 1. Proportion of task-related neurons among different sarnatc~topicallyorganized pop~alatio~zs in precentral cortex (5) 66) 41) (2) Areal Recorded cell Tasksize, density , (4) Task(3) 7 related mm2 unrelated Total no./mtn2 related Nonwrist (F-E) Shoulder Elbow Wrist (other) Fingers Wrist (F-E)

186 40 60 38 48 39

394 153 113 57 7I I7

580 I93 173 95 119 56

32.1 90.7 34.6 40.0 468.3 49.7

80.0 63.1 43.5 46.4 20.8

2.41 2.70 2.18 2.56 2.63

Total

925

41 I

636

35.4

254.8

2.50

found evenly distributed on such a surface (Kwan Classification of PrecentraI Neurms,~ In the present study, the classification derives et al. 1978b; Murphy et aI. 1978; Wong et al. 1978). f r o n ~previous findings that in the precentral forelimb (2) The imumber of cells studied iia each population was proportional to the areal size of the correspondregion cortical neurons have a tight input-output relationship with peripheral forelitlab parts, and ing cortical zone (colul-mn 6, Table 4 ). However, hence can be classified into diRerent populiati~~ns slight departure from uniform sampling was unaccording to single joints (Rosen and Asa~aurna1972; avoidable. This was attributed to the presence of Kwan et al. 1978b; Murphy et al. 1978; Wong et al. blood vessels, which in a few instances precluded any 1978). Other methods of classification of precentral form of neuronal recordings, Another source sf neurons can also be used, including identifying var- sampling bias is related to the electrode impedances. ious populations accordilag to their anatomical con- In general, larger asernrcsmns were preferentially reraectioals within the CNS, nan~e1-ypyramidal-tract corded and studied (Murphy et al. % 978). Elswever, neurons, nonpyraasidal-tract neurons, and other such sampling bias wcsuld presumably occur in every groups sf corticofugal neurons (Evarts 1964; Hum- electrode penetration throughoat the cortical region investigated, and therefore would not prevent a @omphrey and Corrie 1978). parison between functional properties of the spatially Sampling Problem defi~aedpopulations of neurons. In order to establisE~the proportion of tnnits having certain characteristics in each of a ~~kmrxnberof Definition of t8ze Task The task adopted in the present study consisted identified populatiomns, thc problem of sampling bias must be considered. 111the narrower sense, a result of two parts. Tlae first part was designated as the in the context sf the question posed might be con- preliminary period for tIae purpose of control of besidered rneanirmgful only if every neuron in the cor- haviour and brain state, and the second gal% was tical area of itatcrest were recc~rded and studied. basically a stimulus-respcslnse task. In this reaction However, this possibility is not technically feasible. task, mcsrakeys were trained to reposition a handle \Ve adopted an alternative approach to this problem (resgotase) imrmediately after the handle was disby sampling uniformly and cxterrsively to the degree placed by a constant torque perturbation (stimulus). The preliminary period was incorporated in the permitted by present rnicrc~e%ectrcsde recording teclnniques and the anatomy of tlse region, 7 % ~uniform- task for the foE1owing reason. It minimized the variity of sampling should, in prinsiple, allow a mean- ability of the set of the monkeys from trial to trial. ingful iznterpretatio~a of results based on relative The set of the monkey included a proper fixation of proportion of neurons studied. both shoulder and elbow joints, a firm grip on the In the present study, an attempt was made to lakindle by the fingers, and a steady fixation sf wrist sample neurons uniformly throughout the entire joint in its natural position (Stetsora and h%cDill forelimb area by evenly spacing the recording elec- 4923 ) . This preliminary period is important as it has trode penetrations (Wong et a]. 1977; Kwan et al. been demonstrated that the set of the monkey aEects 1978a). Evidence that uniformity of sampling was both motor behaviours and precentral cortical reachieved is reflected in the following observations: sponses in a torque--triggered reaction task ( E v u t s (IB \%'hen data were recorded upon an unfolded f 973). cortical surface, the density of recording sites was During the stinaulus-response period, the delivery

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by WA STATE UNIV LIBRARIES on 11/20/14 For personal use only.

182

CAN. J. PHYSPOL. PHARMACOL. VOL. 57, 1979

of the torque to the handle evoked a sequence of changes in the joint positions which were reflectecl in tlae activities of a specific group of functionally related mrascles. These claanges were followed closely by a compei~satoryreaction of the same group of musclcs to restore the handle. Shortly after the torque oraset, the wrist agonist and antagonist muscles would stretch and relax, sespective%y.In addition, a change in the grippirag pressure exerted by the fingers on the handle and a postural deviation of both elbow joint and shotrlder joint, to a lesser extent from their equilibrium positions, would also occur (Conrad et al. 1975; Smitln et al. 1975). The compensatory rcaction that followed consisted of reflex and voluntary contractio~aof the wrist agonist muscles and other related muscIes in the restoration of the handle, and the adjustment of the handle gripping as well as shoulder-elbow fixation, respectively. All of thc above muscular activities were observed irs the EhfG recordings of this study. In particular, the EMG activities of the wrist agonist muscles observed ira this study are in agreelaaeaat with a previous communication (Tatton and Bruce 1976). F~artherrnore,these observations of EMG activities establish the important precept that solitary contraction of a single n~uscleis a theoretical possibility, but a practical rarity in ordinary nnovernent repertoire. Even in an effort to control firing of a single muscle, a simultaneous inhibition of other functionally related muscles takes place. Critcric~nof Defining Respcansiveness of a Cell T o distinguish objectively a responsive neuron from a nonresponsive one, a criterion based on statistics was sought. Raw data from many repeated trials of a ~pecifictask condition were collected and surnrnated to form a 10-rns bin PSTH. From this BS'FH, values of mean and standard deviation of the firing in the control period were calculated, and then the post-stinaaulus activities were compared witla an arbitrary indicator, which was set at a firing level of two standard deviations from the mean for two or more consecutive bins, in order to determine a significant response. The above test was estimated to accept or to reject cortical responses at a significant level of p < 0.0885. Tests of similar form to detect significant responses in the precentral cortex, based on the same statistical principle, were formulated in two previous studies (Thach 1975; Tanji and Evarts 1976). In both studies, PSTHs of 2-ms bin width were chosen for the purpose. However, two different levels of significance were used: g < 0.081 in one study ('Fa~ajiand Evarts 1976) and p < O.OOO0I in the other (Thach 1975).

brnpltcuticzns of Exgerimenta! Results Withirn the limitations ianposcd by methoc%olc~gica consideratioras discussed in previous sections, the data iaa Table 1 are interpreted as follows: ( 8 ) Both wrist (F-E) netarcxu and ncsnwrist (F-E) neurons are active in the task. (2) Not aIT wrist (F-E) neurons are related to the task. (3) The wrist (F-E) neurons are predominantly active in comparison evitla other populations of precentral neurons. The first firading, that both wrist (F-E) and msnwrist (F-E) neurons are active in a task primarily involving wrist F-E msvements, supports the hypothesis stated in the Introduction, that a correlation exists between precentral ineuronal activities and peripheral forelimb ~novcrnents i~mvolvingthe lnomor-aymous jcdints. Thus, a populatioia of wrist neurons principally controls outputs acting on the wrist joint and receives inputs from its solnaatic surround, with corresponding relationships for finger, elbow, and shoulder neurons in the precentral cortex. I n the case of the present task and similarly of other tasks in which there is a cocsrdinated movement about the wrist joint in the F-E plane together with other functionally related forelimb joi~~fs, one would expect the following neuronal activity patterns: shoulder-elbow neurons in postural fixation of forearnn, finger neurons in handle gripping, and wrist (F-E) and wrist (other) neurons in repositioning of the handlc. Results similar to the above finding were also observed in previous studies in which precentral neurons were identified by intracortical micrc~stimulation (Colasad et al. 1975 ; Smith el, af. 1975; Ysamiya 1977). In these reports, while monkeys were performing a task involving movements about a foreli~nbjoint, cortical cells related to the task were found inside and outside a localized prece~ata-a1cortical region that was identified to represent this forelimb joint. It is interesting to note that other functional frarncworks for precentral neurons can be identified and demonstrated experimentally, for example control of motoneuron pools (Preston and Whitlock % 96 1; Eandgren et al. 6962) and single muscles (Clough et al. 197%; Fetz and Fii~occhio %972).Tkcse frameworks may be considered subsets of the present one, which is based on the physiological goal of the system, namely movement. In relation to the secornd finding that not all wrist (F-E) neurons are related to the task, a similar result was reported previously (Yurniya 1977). These data indicate the important finding that it is highly probable that only selected neurons from functionaBly identified groups of neurons in precentraI cortex are involved in a specific motor task. The

WONG ET AE.

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by WA STATE UNIV LIBRARIES on 11/20/14 For personal use only.

result would have three implications. Firstly, other functionally similar neurons froan such a group would be saved for other tasks. Secondly, the functional redundancy which may be necessary to recovery of motor skills after partial lesions (Grbinbaum aand Sherrington 1903; Glees and Coles 1950) would be provided. Thirdly, the functional substrate for plasticity development of motor skills (Hinee; 1943) would be embodied. The third and final finding of this study, that the wrist (F-E) neurons are predominantly active in comparison with other populatioils of neurons, adds another line of evidence in support of the idea of a correlation between peripheral forelimb movements and precentral neuronal activities. That amonw~st (F-E) populations, in particular the shoulder population, are related to the task to a lesser cxtent is consistent with the primarily supportive role that the measclles which their neurons co~atrolplay during this task. Based on these findings, one is led to speculate that in a task primarily ii~volvingmovement about another forelimb joint, say the elbow joint, a significantly higher proportion of elbow neurons as compared with other populations would be task related. In co~aclusian,the findings of this study support the major principle that representation of peripheral events in precentral cortex is embodied in terms of the ensemble behaviour of separate neursnal populations, which control movement about single joints. This feature of brain-body interaction may be utilized irm the execution of learned movements.

Acknowledgments We wish to express our gratitude to Mr. H. Ngsnyen-Huu for his computer expertise in the control of experiments and data acquisitioim. This work was supported by the Medical Research Council of Canada, grant No. MT-42 40. ASANUMA: H.. and SAKATA. H. 1967. Fr~nctionalorganization of a cortical efferent system exanrined with focal depth stimulation in cats. J. Neurophysiol. 40,35-54. CL~SWGH, J. F. M., PHILLIPS, C. G., and SHERIDAN, 3. D. 1971. The short-latency projection from the baboon's motor cortex to fusimotor neurons of the forearm and hand. J. Physiol. (London), 214: 257-2914. CONRAD,B., MEYER-LCIHMANN, J., MATGUNAMY, K., and BROOKS,V. B. 3975. P r ~ e n t r a lunit activity following torque pulse injections into elbow movetnents. Brain Res. 94, 219-236. EVARTS, E. V. 1964. Temporal patterns of discharge of pyramidal tract neurons during sleep and waking in the monkey. J. Neurophysiol. 27, 152-1 7 1. 1966. Pyramidal tract activity associated with a condi-

183

tioned hand movement in the monkey. J. Neurophysiol. 29, 1011-1027. 1973. Motor cortex reflexes associated with learned movement. Science, 179, 501-503. FETZ,E. E., and ~ 1 ~ 0 ~ ~ 1D.1 V. 1 01972. , Operant conditioning of iwlated activity in specific muscles and precentral cells. Brain Res. 40, 19-24. GLEES,P., and C o ~ e s ,J. 1950. Recovery of skilled motor functions after small repeated lesions of motor cortex in macaque. J. Neurophysiol. 13, 3 37-148. GRUNBAUM, A, S. F., and SHEKRINGTON, C. S. 1903. Observations OII the physiology of the cerebral cortex s f the anthropoid apes. Prsc. R. Soc. London Ser. B, 72, 152-155. HINES,M. 1943. Control of movenaents by the cerebral cortex in primates. Biol. Rev. 18, 1-31. HUMPHREY, D. W., and CORRIE,W. S. 1978. Properties of pyramidal tract neuron system within a functionally defined subregion of primate motor cortex. J. Neurophysiol. 41, 2 B 6-243. JASPER, H. H. 1958. Recent advances in our understanding of ascending activities of the reticular system. b/n Reticular formation of the brain. Edited by H. H. Jasper. Little, Brown & Co. Ins., Boston. pp. 423-434. KWAN,M. C., MAGKAY, W. A., MURPPIY, J. T., and WONG, Y. C. 1978a. An intracortical microstimulation study of output organization in precentral cortex of awake primates. J. BPhysiol. (Paris), 74, 231-233. 1978b. Spatial organization s f precentral cortex in awake primates. 11. Motor outputs. J. Neurophysiol. 41, 1120-1131. LANDGREN, So,~%ILLIPS, C. G., and PORTER, R. 1962. Minimal synaptic actions of pyramidal inlpulses on some alpha nnotoneurones of the baboon's hand and forearm. I. Physiol. (London), 161, 91-1 11 1. MURPHY, 3. T., KWAN,H. C., MAGICAY, W. A., and WONG, Y. C. 1974. Evaluation of neuronal spike trains in neurophysiological experiments. Physiol. Behav. 13, 3 13-31 5. 1978. Spatial srga~aimtion of precentral cortex in awake primates. III. Input-output coupling. J. Neurophysiol. 48, 1132-1 1139. O'RIAIN,M. Do,BLAIR,W. D. G., and MURPHY, J. T. 1978. The influence of a visual cue on muscle stretch reflexes. Can. J. Physiol. PharmacoB. 56, 19-22. PRESTON, J. B., and WHITLOCK,D. G. 1961. Bntracellular potentials recorded from motoneuresns following precentral gyrus stimulation in primate. J. Neurophysiol. 24, 91-180. ROSEN,I., and ASANUMA, M. 1972. Peripheral afferent inputs to the forelimb area of the monkey motor-cortex: inputoutput relations. Exp. Brain Wes. 14, 259-273. SNINODA, Y.,ARNOLD,A. P., and ASANUMA, H. 1376. Spinal branching of corticospinal axons in the cat. Exp. Brain Res. 26, 215-234. SMITH, A. M., HEPP-REYMOND, M. C., and Wyss, %I. R. 6975. Relation of activity in precentral cortical neurons to force and sate of force change during isometric contractions of finger. Exp. Brain Res. 23, 315-332. STETSON, R. H., and MCDILL,3. A. 1923. Mechanisms of the different types of anovernents. Rychol. Monogr. 32, 18-40. TANJI,J., and EVARTS, 13. $1. 1976. A~lticipatoryactivity of motor cortex iaeurons in relation to direction of an intended movement. J. Neurophysiol. 39, 1062-8068. TATTON,Mr. G., and BRUCE,1: C . 1976. Temporal relations between motor cortical ~aeuronal responses and EMG responses to sudden load changes applied to the primate wrist. Can. Physiol. 7, 59.

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by WA STATE UNIV LIBRARIES on 11/20/14 For personal use only.

184

CAN. J. PEHYSHOE. PWARMACOE. VOL. 57, 1978

THACH, W. T. 1975. Timing of activity in cerebellar dentate 1978. Spatial organieation of prccentral cortex in nucleus and serebrai motor cortex during prompt volitional awake primates. I. Sealory inputs. J . Idenrophysiol. 41, rtlovernerat. Brain Res. 88, 333 -24 1. 1107--I119. WC)NG~ Y . C.,K V V A NH. ~ C,, MACK~ZEI, W. A., and Mh:ra~~rn., YUMIYA, H. 1977. Neuronal activity in cortical efTercnt zones J. 'T.1977. Topogrraphic organization of afferent inputs in projecting t s wrist extcnsors drlring volurltary wrist extenmonkey precentral cortex. Brain Res. 138, 166-1 68. sion in the monkey. Tohoku J . Fxp. Med. 121, 321-326.