Chapter 28 Eberhard E. Fetz

Motor Functions ofCerebral Cortex

IntroductionPrimary and Secondary Motor Cortex AreasThe Pyramidal and Extrapyramidal Systems

Organization of Pyramidal Tract NeuronsPrimary Motor Cortex ( MI)

Somatotopic OrganizationEffects of Stimulation on MusclesEffects of Stimulation on MotoneuronsIntracortical Microstimulation

Cortical ConnectionsCortical Neurons and Their TargetsSources of Inputs to Primary Motor CortexSomatic Sensory Input to Motor CortexInput-Output Relations

Coding of Movement Parameters by MotorCortex Cells

Relative Timing of Cell Activity: Reaction TimeResponses

Relation to Active ForceLoad Compensation ResponsesPreparatory SetCoding of Movement Parameters by

Population ResponsesEffects of Motor Cortex Lesions

Spasticity and ParalysisEffects of Pyramidal Tract Lesions

Motor Functions of Other Cortical-AreasSupplementary Motor CortexPremotor CortexFrontal Eye FieldsPosterior Parietal Cortex

Distributed Cortical Motor Function

INTRODUCTION

Primary and Secondary Motor Cortex Areas

Anyone who has seen the victim of a cerebralstroke can appreciate the importance of motorcortex in performing normal movements. Typi-cally, such patients are incapable of using one or

more contralateral limbs and frequently cannotspeak fluently. The loss of control is most notablein the muscles of the hands and face. Furthermore,many patients are unable to regain normal controlof the paralyzed muscles, indicating that otherregions cannot compensate to perform these motorfunctions.

Clinical and experimental evidence indicatesthat the cerebral cortex plays a crucial role in boththe programming and the execution of normalvoluntary movements. The relevant cortical areascan be broadly divided into two groups: primarymotor cortex, which has relatively direct anatom-ical and functional relation to muscles and isimportant for the normal execution of movements,and secondary motor cortical areas that are synap-tically more remote from the periphery and aremore involved in programming movements underparticular circumstances. The locations of thesecortical regions in monkeys and humans areshown in Figure 28-1. The largest of these areas,the primary motor cortex (MI) lies on the precentralgyms, in Brodmann's area 4. Electrical stimulation.of the primary motor cortex produces specific and-repeatable movements, with the lowest stimulusthresholds of any cortical area. MI contains a'somatotopic map of the representation of bodymuscles. Electrical stimulation of secondary areasgenerates movements more rarely; these move-ments are more complex and variable and requirestronger stimulus currents. These secondary mo-tor areas include the premotor cortex (PMC), whichlies anterior to MI, in the lateral portion of Brod-mann's area 6, and the supplementary motor cortex,or supplementary motor area (SMA), which liesin the medial portion of area 6, largely in themedial bank of the sagittal sulcus. Both of theseregions are also somatotopically organized and areinterconnected with each other and MI. Still fur-ther rostral, in Brodmann's area 8, are the frontaleye fields, which are concerned with movements of

Figure 28-1 Major motor areas of cerebral cortex in (A)macaque monkey and (B) man. Numbers refer to Brodmann'sdesignation of cortical regions with different cytoarchitectonicfeatures. (From Brodmann, K Vergleichende Lokalisationslehreder Grosshirnrinde. Leipzig, Barth, 1909.)

the eyes. Finally, the posterior parietal cortex,' com-prising Brodmann's areas 5 and 7, appears to beinvolved in programming directed movements ofthe limbs and eyes to targets in space.

In addition to electrical stimulation, other linesof experimental evidence support a functional dis-tinction between primary and secondary motorcortical areas. Lesions in primary motor cortex ofprimates tend to produce paresis or paralysis,obvious deficits in generating muscle activity. Le-sions in secondary motor regions produce moresubtle deficits, called apraxias, which are the in-ability to perform certain types of movementsunder particular conditions. Finally, the recordingof neuronal activity during movements has revealedthat many primary motor cortex cells discharge

strongly in relation to muscle activity, whereasneurons in secondary motor areas tend to be activeunder particular conditions that give rise to move-ments. This chapter reviews the evidence thatprimary motor cortex is predominantly involvedin the execution of movements whereas secondarymotor areas are more concerned with motor pro-gramming. The major emphasis is on primarymotor cortex, which has been studied most exten-sively and is best understood.

The Pyramidal and Extrapyramidal Systems

The cerebral cortex may influence movementsthrough two major output pathways: the pyrami-dal and the extrapyramidal systems. While thesetwo systems normally work together to controlmovement and posture, their functions are distin-guished by neurologists who classify motor defi-cits as "pyramidal" or "extrapyramidal." The py-ramidal system consists of cortical neurons whoseaxons descend through the medullary pyramids;the vast majority continue to the spinal cord viathe corticospinal tracts, but some terminate in thebrain stem reticular formation. The extrapyramidalsystem comprises the remaining motor systems,including the descending tracts from brain stemto spinal cord, the basal ganglia and the cerebel-lum. Inclusion of many different centers in theextrapyramidal system limits the usefulness of thisclassification, as discussed in Chapter 26.

Organization of Pyramidal Tract Neurons

The corticospinal tract exists only in mammalsand increases in size along the phylogeneticscale. 24' 36a In humans, each pyramid containsabout one million fibers. About 60% of the pyram-idal tract arises from cells in precentral cortex(approximately 30% from Brodmann's area 4 and30% from area 6); the remaining 40% of the pyram-idal tract arises from postcentral cortex. Figure 28-2 illustrates the course of precentral and postcen-tral pyramidal tract fibers in the macaque. Theprecentral pyramidal tract neurons originate inboth primary and secondary motor areas and ter-minate preferentially in the more ventral laminaeof the cord, which contain motoneurons and in-terneurons involved in movement. The numberand proportion of corticospinal fibers terminatingamong motoneurons becomes progressivelygreater from prosimians, through monkeys andapes, to humans. Postcentral pyramidal tract cellstend to terminate more dorsally in the spinal cord,

Figure 28-3 Cortical location ofpyramidal tract neurons labeledby retrograde transport of horse-radish peroxidase (HRP) from con-tralateral thoracic spinal cord ofthe cynomolgus monkey. A, Lo-cation of labeled cells projectedon flattened cortical surface. B,Representative sagittal sectionthrough cortex at arrow in A. Num-bers denote Brodmann's areas.(From Jones, E. G.; Wise, S. P. J.Comp. Neurol. 175:391-438, 1977.)

610 q V CONTROL OF MOVEMENT

Figure 28-2 Course and termination of pyramidal tract fibersfrom precentral and postcentral cortex of the monkey. Heavydots and line segments show the course of fibers descendingin lateral and ventral corticospinal tracts. Light dots show sitesof termination of corticospinal fibers. (From Kuypers, H. G. J.Brain 83:161-184, 1960.)

among interneurons relaying peripheral inputfrom afferent fibers to motoneurons and to highercenters. This postcentral component of the pyram-idal tract may be more involved in regulatingsensory function by presynaptic and postsynapticmodulation of transmission of afferent impulses.

The cortical origin of the pyramidal tract hasbeen documented most clearly by retrograde trans-port of horseradish peroxidase (HRP) from thespinal cord or the pyramids. In the monkey, cor-ticospinal cells labeled with HRP are extensivelydistributed over several cortical fields, includingBrodmann's areas 8, 6, and 4 precentrally, andareas 3, 1, 2, 5, and 7 postcentrally (Fig. 28-3A).Sagittal sections of the cerebral cortex reveal thatthe somata of pyramidal tract neurons all residein layer 5 of the cortex (Fig. 28-3B). The labeledcorticospinal cells appear to be grouped in clusters,which on surface reconstructions tend to formmediolateral bands. 18

Most of the pyramidal tract fibers are small-90% are less than 4 um in diameter—and abouthalf are unmyelinated. Thus the majority of py-ramidal tract fibers are slowly conducting. Thelarge Betz cells in motor cortex contribute about2% of all pyramidal tract fibers. In the primate,single pyramidal tract cells typically distribute ter-minals to many motoneuron pools. 31

PRIMARY MOTOR CORTEX (MI)

Somatotopic Organization

Effects of Stimulation on Muscles

The cortical motor areas have been delimitedand mapped by observing the movements evokedby electrically stimulating the cortex. The primarymotor cortex (MI), which has the lowest thresholdsites, is organized somatotopically much like the

primary somatosensory cortex. Thus stimulatingadjacent cortical sites evokes movements of adja-cent body parts. The representation of the somaticmusculature in MI of the monkey is summarizedin Figure 28-4A. The motor map in the precentralgyrus provides the most extensive and detailed

cortical representation of limb muscles. The ana-tomical distortion of the simunculus figure (Fig.28-4A) indicates that a proportionately larger re-gion of cortex is devoted to those body partscapable of finer motor control, such as the tongue,digits, and toes. The greater cortical area devoted

Figure 28-4 Somatotopic organization of motor cortical areas in monkeys and humans. A, The summary of body regionsactivated by cortical surface stimulation in an anesthetized macaque monkey. MI = primary motor cortex; SMA =supplementary motor cortex; SI = primary somatosensory cortex; SII = secondary sensory cortex. B, Relative size and locationof primary motor cortex areas devoted to different body parts in humans. Sensory input is shown at left; motor outputrepresentation is shown at right. (A, After Woolsey, C. N. In Harlow, H. F.; Woolsey, C. N. Biological and Biochemical Bases ofBehavior, Madison, Wis., Univ. of Wisconsin Press, 1958; B, reprinted with permission of Macmillan Publishing Company fromPenfield, W.; Rasmussen, A. T. Cerebral Cortex of Man. A Clinical Study of Localization of Function. Copyright © 1950 byMacmillan Publishing Company, renewed 1978 by Theodore Rasmussen.)

to these regions contains a larger number of inter-neurons concerned with controlling this outputand also provides more efferent cells that projectto the brain stem and spinal cord.

Many of the fine movements evoked by motorcortex stimulation in the monkey are mediated bythe pyramidal tract. This is illustrated by experi-ments in which the pyramidal tract was sectioned.Figure 28-5 summarizes the results of an experi-ment in which Woolsey and colleagues' sectionedthe pyramidal tract unilaterally in monkeys andthen mapped the movements evoked by stimulat-ing the cortex on both sides. Movements elicitedfrom the normal side, with the pyramidal tractintact (Fig. 28-5, right), were diverse and com-monly involved distal finger and toe movements.The cortical distribution is in general agreementwith the summary map of Figure 28-4A. Althoughmovements could also be evoked by stimulatingthe cortex with the pyramidal tract sectioned (Fig.28-5, left), these movements were more restrictedand involved mainly proximal joints such as theknee and elbow. Moreover, these movements re-quired stimulus intensities two to three times

higher than those of corresponding points on theside with the intact pyramidal tract. This experi-ment demonstrates that the pyramidal tract isimportant in mediating outputs from motor cortexto motoneurons of distal muscles.

A comparable somatotopic motor map for hu-man cortex was obtained by Penfield and col-leagues, 25 who stimulated the surface electricallyduring neurosurgical procedures to identify thefunctions of cortical sites. As in the monkey, themost medial portions of human motor cortex aredevoted to the leg and the more lateral regions tothe arm and face, as indicated in Figure 28-4B.Again, the relative area of cortex devoted to thethumb and tongue greatly exceeds their relativeanatomical size. Some of this topographic organi-zation of human motor cortex had already beendeduced by the neurologist Hughlings Jackson byobserving the typical progression of epileptic sei-zures. The so-called "Jacksonian march" of epilep-tic activity begins with twitches of the thumb andthen proceeds to involve the hand and moreproximal parts of the body, and eventually theface. Reasoning that these peripheral symptoms

(465 Days)

Figure 28-5 Figurine map illustrating movements evoked by electrical stimulation of pre- and postcentral cortex of macaquemonkey with a unilateral pyramidal tract section. Figurines are located at the cortical points that evoked movements of thejoints indicated. Pyramidal tract from left hemisphere was sectioned and from right hemisphere was largely intact. c.f. -central fissure; i.p.s. = intraparietal sulcus; a.s. = arcuate sulcus. (After Woolsey, C. N.; Gorska, T.; Wetzel, A.; Erickson, T. C., Earls,F. J.; Allman, J. M. Brain Res. 40:119-123, 1972.)

reflect the spread of epileptic activity throughadjacent cortical areas, Jackson deduced much ofthe relative cortical representation of the muscu-lature shown in Figure 28-4B. 25b

Conscious human patients whose precentral gy-rus is electrically stimulated often report that theevoked movements are involuntary. Precentralstimulation evokes simple, stereotyped motor re-sponses, rather than any conscious impulse tomove, indicating that the primary motor cortex isrelatively close to the output of the motor system.At many sites of human motor cortex electricalstimulation produces relaxation of musculaturerather than muscle contraction.

In addition to evoking involuntary movementsand arresting voluntary movements, electricalstimulation of motor cortex may also evoke so-matic sensations. About a third of the precentralcortex sites stimulated by Penfield and Boldrey25

evoked experiences of tingling or numbness—similar to the sensations produced by stimulatingpostcentral cortex. These sensations could beevoked from precentral sites even after ablation ofadjacent postcentral regions that normally processsomatic sensation (see Chap. 15), suggesting thatprecentral cortex may also be involved in somaticsensation.

Because cortical stimulation evokes movementsthat involve many muscles, it has been arguedthat movements rather than muscles are repre-sented in motor cortex. In this view, the primarymotor cortex sends command impulses that signalbrain-stem and segmental circuitry to generatecoordinated activity of many motoneuron pools.The alternative view is that motor cortex cellsultimately control individual muscles, and thattheir activity is coordinated by inputs from highercenters. To test these alternatives, Chang, Ruch,and Wards dissected ankle muscles in anesthetizedmonkeys and measured the tensions produced byrepetitive stimulation of different cortical sites.Each muscle could be made to contract from stim-ulation of a wide region of cortex, but the sites oflowest threshold were more localized. Stimulationof some low-threshold sites evoked contraction ofspecific muscles in isolation, whereas stimulationof other sites caused several muscles to contracttogether, suggesting that MI can affect both singleand multiple muscles.

Effects of Stimulation on Motoneurons

The effects of corticospinal projections havebeen analyzed in more detail by recording intra-cellularly the excitatory postsynaptic potentials

(EPSPs) evoked in motoneurons by cortical stim-ulation.26,27 In primates, some corticospinal cellsmake monosynaptic connections with motoneu-rons. Figure 28-6A illustrates corticomotoneuronalEPSPs recorded in motoneurons innervating armmuscles of the baboon. 26 As the stimulus intensityincreases and recruits more corticomotoneuronalcells, the amplitude of the EPSPs also increases(Fig. 28-6B). Above a certain stimulus intensitythe size of the EPSP increases no further, suggest-ing that the whole "colony" of corticomotoneu-ronal cells projecting to that motoneuron has beenrecruited. To estimate the cortical extent of themotoneuron's colony, Phillips and Porter26 meas-ured the current spread of the cortical stimuli bymeasuring the threshold intensity required toevoke a response in single pyramidal tract cells asa function of distance from the lowest thresholdpoint. The threshold typically increased as thesquare of the distance, as shown in Figure 28-6C.Using this relation, the extent of the cortical colo-nies was estimated to range from 2 to 10 mm 2 fordifferent motoneurons. EPSPs in distal motoneu-rons reached their maximum at lower stimulusintensities than those in proximal motoneurons,suggesting that the cortical colonies of distal mo-toneurons were more localized in the cortex. More-over, motoneurons of distal muscles also receivedlarger maximal EPSPs than those of proximal mus-cles. Thus, motoneurons of distal muscles receiveda greater net synaptic input from cortex, and thisoriginated from a smaller cortical area.

The spatial distribution of cortical colonies wasdetermined by mapping the cortical points fromwhich minimal corticomotoneuronal EPSPs couldbe evoked in hindlimb motoneurons of the mon-key.16 The cortical distributions of these colonieswere irregular in shape, typically between 3 and7 mm2 , and the colonies projecting to motoneuronsof different muscles overlapped extensively. Dif-ferent motoneurons of the same pool sometimesreceived monosynaptic EPSPs from different cor-tical areas, suggesting that some corticomotoneu-ronal cells may project to specific motoneuronswithin the pool. Cortical stimulation also evokeddisynaptic inhibitory postsynaptic potentials(IPSPs) in motoneurons; these inhibitory effectsare mediated in part by the Ia inhibitory interneu-rons, which receive monosynaptic input from thecortex.

The relative magnitudes of the monosynapticinputs to motoneurons descending from motorcortex and those arriving from Ia muscle spindleafferents suggest that the cortical and reflex inputsare somewhat different for each muscle. The size

Figure 28-6 Effects of cortical stimu-lation. A, corticomotoneuronal excita-tory postsynaptic potentials (CM-EPSPs)evoked in a motoneuron by increasingintensity of cortical surface stimulation.8, Size of CM-EPSP as function of stim-ulus intensity for motoneuron of distaland proximal muscle. C, Spread of cor-tical surface stimulus as measured bythe threshold intensity required to ac-tivate pyramidal tract cell; typicalcurve of threshold in milliamperes, asa function of distance (D) of the stimu-lating electrode (S) from the best point(for 0.2 ms anodal pulse). D, Spread ofintracortical microstimulation as meas-ured by threshold intensity required toactivate cortical cell; typical curve of

threshold in microamperes, as a function of distance (d) of the stimulating electrode (S) from the best point (for 0.2 ms cathodalpulse). R represents recording electrode. (A, B, and C from Phillips, C. G.; Porter, R. Prog. Brain Res. 12:222-242, 1964 [8 and Cmodified]. D after Stoney, S. D.; Thompson, W. D.; Asanuma, H. J. Neurophysiol. 31:659-669, 1968.)

of the maximal EPSP from these two sources isrepresented in Figure 28-7A by the width of thearrows converging onto motoneurons innervatinghand and finger muscles of the baboon. The ex-tensor digitorum communis (EDC) motoneuronsreceive the largest corticomotoneuronal EPSPs,which exceed the magnitude of the Ia-EPSPs fromEDC muscle afferents. Other motoneurons, suchas those of the hypothenar muscles (Uh), receivegreater net synaptic input from muscle spindlesthan from cortical inputs.

Not only does electrical stimulation of MI pro-duce monosynaptic EPSPs in primate motoneu-rons, but repetitive cortical stimulation, whichmimics the bursts of activity recorded in cortico-spinal cells, enhances the effectiveness of EPSPsbeyond simple summation. Figure 28-7B showsthat repetitive stimulation (200/s) of the Ia muscleafferent produced algebraic summation of succes-sive Ia-EPSPs, whereas repetitive stimulation ofmotor cortex produced an increase in the size ofsuccessive corticomotoneuronal EPSPs. This in-crease was not due simply to recruitment of ad-ditional corticospinal cells, since the amplitudes ofthe descending volleys remained constant. Neithercould the increase be due to recruitment of exci-tatory spinal interneurons, since the latency andshape of the EPSPs did not change, but only their

amplitude. Such facilitation of corticomotoneu-ronal EPSPs enhances the effectiveness of high-frequency activity, which typically occurs at theonset of movement (see Fig. 28-14).

Intracortical Microstimulation

The experiments described above employed cor-tical surface stimulation, which can spread overconsiderable distances, limiting the spatial reso-lution of the effective sites. To evoke discreteresponses from more circumscribed intracorticalsites, Asanuma and Rosen' delivered intracorticalmicrostimulation through microelectrodes. Thistechnique has provided three-dimensional mapsof low-threshold output sites in the depth of thecortex. To quantify the effective spread of stimuluscurrents, Asanuma and colleagues 3la measured thestimulus intensities required to activate a pyrami-dal tract neuron as a function of the distance ofthe stimulating electrode. As the electrode wasmoved from the lowest threshold point, thresholdintensities increased as the square of the distance(see Fig. 28-6D). The effective radius of a 10-uacurrent pulse was found to be 80 to 90 pm.

Using intracortical microstimulation to evokeactivity in a given muscle, Asanuma and Rosen'found that the low-threshold points for individual

la

200/IN

CM5

mV

B -

muscles were located in a cylindrical volume ofcortex perpendicular to the cortical surface (seeFig. 28-12). The diameters of these columnar zonesranged from 0.5 to several millimeters. These ob-servations suggested a columnar organization ofoutput cells to specific muscles, analogous to thecolumnar arrangement in sensory cortex of cellswith inputs from particular receptors. However,in addition to activating cells directly, repetitivemicrostimulation is very effective in evoking re-sponses transsynaptically, via fibers that are acti-vated directly. This could explain why the low-threshold points for evoking muscle responseswith repetitive stimulation have the same distri-bution as the afferent and efferent fibers. Thesefibers could activate the corticospinal output cells,which are located in cortical layer V (see Fig. 28-3B).

Cortical Connections

Cortical Neurons and Their Targets

The operations of the motor cortex can best beunderstood in the context of its input and outputconnections. On the basis of cell morphology andfiber distributions, anatomists have distinguishedfive layers within motor cortex. Unlike other cor-tical areas, primary motor cortex lacks a prominentlayer IV containing granule cells, so the precentralgyrus is also called "agranular cortex." Drastically

simplified, cortical cells can be classified into twobasic types: pyramidal and nonpyramidal. Pyram-idal cells, so called because of their pyramidalsomata, have a long apical dendrite directed to-ward the cortical surface and short basal dendritesissuing from the base of the soma. Their extensivedendritic tree suggests that they integrate synapticinput from several layers. Axons of the pyramidalcells leave the cortex for other cortical and subcor-tical targets, making them the main cortical outputcell; in addition, axon collaterals of pyramidal cellsusually provide local intrinsic connections. Non-pyramidal cells, on the other hand, have smalldendritic trees that sample input from a morerestricted region; because their axons remainwithin the cortex and arborize locally, they areinterneurons. Nonpyramidal cells include stellate,basket, and granule cells, some of which are in-hibitory.

As illustrated in Figure 28-8, pyramidal cells ineach cortical layer send axons to specific targets.The pyramidal cells in layers II and III project toother cortical regions. The more superficial cellsproject to ipsilateral cortical areas, notably to sec-ondary motor areas (SMA and PMC) and to post-central sensory cortex; the deeper cells in layer IIIsend axons through the corpus callosum to themotor cortex of the opposite hemisphere. Thesetwo superficial cortical layers in turn receive theirinput from other cortical regions. Thus, layers IIand III are largely concerned with intercortical

Figure 28-7 Comparison of monosynaptic excitatory postsynaptic potentials (EPSPs) evoked in forelimb motoneurons fromcerebral cortex (CM) and from la afferent fibers. A, Total synaptic input to different types of motoneurons. Thickness of arrowsrepresents the relative size of maximal EPSP. EDC = extensor digitorum communis; R = remaining dorsiflexors of wrist; Uh =intrinsic hand muscles innervated by ulnar nerve; Mh = intrinsic hand muscles innervated by median nerve; FDS = flexordigitorum sublimis; PL = palmaris longus. B, High-frequency stimulation evokes facilitation of successive corticomotoneuronal(CM) EPSPs (bottom) but not la EPSPs (top). Upper records show that arriving volleys, recorded on cord dorsum, are constant.(A from Clough, J. F. M.; Kernell, D.; Phillips, C. G. J. Physiol. [Lond.] 198:145-166, 1969. B, from Phillips, C. G.; Porter, R. Prog. BrainRes. 12:222-242, 1964.)

communication between somatotopically relatedareas.

Most of the outputs to subcortical targets arisefrom pyramidal cells in layer V. The corticospinalcells lie deepest in layer V and include the largestpyramidal cells in motor cortex, the so-called Betzcells. Cells in successively more superficial por-tions of layer V project to successively more prox-imal brain-stem targets, namely to the medulla,pons, and red nucleus; the most superficial layerV cells project to the striatum. A few pyramidalcells send branching connections to more than oneof these targets, but the majority project to onlyone site.

Layer VI contains smaller pyramidal cells withlong apical dendrites, whose axons project to thethalamus, particularly the ventrolateral nucleus.Many layer VI cells also send recurrent connec-tions to upper cortical layers.

Sources of Inputs to Primary Motor Cortex

The major inputs to primary motor cortex (MI)arise from other cortical areas and from thalamicnuclei. Interconnections between ipsilateral corticalregions are diagrammed in Figure 28-9. These

cortical areas are all reciprocally connected in asomatotopically organized manner. The most mas-sive connections are those that interconnect ho-mologous pre- and postcentral cortex sites. Theseinputs to motor cortex provide cutaneous andproprioceptive somatosensory information (andmotor cortex, in turn, provides information aboutmotor commands, which can modulate the activityof sensory cortex cells). However, postcentral cor-tex is not the only source of sensory input, sincesomatic stimuli can still evoke responses in motorcortex after ablation of postcentral cortex. MI isalso reciprocally connected with the SMA andPMC. As discussed below, these secondary motorareas are commonly considered to be sources ofmotor commands to primary motor cortex. Affer-ent connections from other cortical areas convergein the superficial layers, among the cells thatproject back to the same cortical regions.

Contralateral motor cortical regions are connectedthrough the corpus callosum. The corpus callosumpreferentially interconnects those regions relatedto limb girdle and axial musculature (face andtrunk regions), suggesting that it helps to integratemotor control of these midline structures. Thehand and foot areas of motor cortex (like those of

Figure 28-9 Major interconnections between cortical areasinvolved in motor control. Numbers indicate Brodmann's areasand arrows indicate direction of projection. Regions arereciprocally interconnected in a somatotopic manner. SMA= supplementary motor area. (After Wiesendanger, M. InTowe, A. L.; Luschei, E. S., eds, Handbook of Behavioral Neuro-physiology, vol. 5, pp 401-491. New York, Plenum Press, 1981.)

sensory cortex) are not interconnected with thecontralateral side. The supplementary and pre-motor cortex areas are also interconnected withtheir contralateral counterparts. Surprisingly, sec-tioning the corpus callosum does not noticeablydisrupt motor coordination of proximal or distallimb muscles.

The main thalamic input to primary motor cortexcomes from the ventrobasal nucleus (nucleus ven-tralis lateralis caudalis [VLc] and nucleus ventralisposterolateralis oralis [VPLo], using the nomencla-ture of Olszewski23), which relays informationabout peripheral events arriving from the spino-thalamic tract as well as central commands arrivingfrom the cerebellar nuclei. Recent evidence revealsthat the primary and secondary motor corticalareas each receive their thalamic inputs from sep-arate nuclei, which in turn transmit inputs fromdifferent sources. As schematized in Figure 28-10,the VPLo and VLc nuclei relay impulses from therostral dentate nucleus and from spinothalamiccells to primary motor cortex; nucleus ventralislateralis oralis (VLo) transmits input from the basalganglia (globus pallidus and substantia nigra) tothe supplementary motor cortex, and nucleus Xrelays information from the caudal dentate nucleusto the lateral premotor cortex. The existence ofseparate subcortical input pathways to each of thethree motor areas implies that the basal gangliaand caudal dentate nucleus could influence pri-

mary motor cortex only indirectly, via relaysthrough their secondary motor fields.

In addition to these "specific" motor relay nuclei, thethalamus also provides input to motor cortex from"nonspecific" nuclei such as the intralaminar and retic-ular nuclei; these are thought to regulate the generalexcitability of cortical neurons (Chap. 15).

Somatic Sensory Input to Motor Cortex

Most cells in motor cortex receive somatic input,which can be readily demonstrated by their re-sponses to peripheral stimulation. In the unanes-thetized primate, two thirds of the MI cells re-spond clearly and consistently to adequate somaticstimulation. Most cells can be driven by passivejoint rotation, usually flexion or extension of oneor more joints. Other motor cortex cells respondto cutaneous stimulation, such as touching skin orbrushing hairs; their receptive fields are similar insize to those of postcentral cortex cells. A fewprecentral cells receive both deep and cutaneousinput. About a third of the precentral neuronsshow no clear somatosensory responses, althougha few respond to visual or auditory stimuli.

Cells encountered in a vertical electrode pene-tration through the motor cortex tend to respondto input of the same modality (deep or cutaneous)and to receive input from the same joint or over-lapping receptive fields. This tendency for motor

Figure 28-10 Schematic sag-ittal section through pre- andpostcentral cortex, showingBrodmann's cytoarchitectonicareas and their thalamic inputs.The predominant sensory mo-dality represented in each areais symbolized by C (cutaneous),P (proprioceptive), or V (visual).Note the separate thalamocort-ical pathways to the three motorcortical areas. Primary motorcortex (MI) is connected recip-rocally with ventralis posterola-teralis oralis (VPLo) and ventralislateralis caudalis (VLc); supple-mentary motor area (SMA), withventralis lateralis oralis (VLo); andpremotor cortex (PMC), witharea X. Input from cerebellum isrelayed via the caudal and ros-tral components of the dentatenucleus (DNc and DNr, respec-tively). (After Jones, E. G. In Jones,E. G.; Peters, A, eds. CerebralCortex, vol. 5,113-114. New York,Plenum Press, 1986. Schell, G. R.;Strick, P. L J. Neurosci. 4:539-560,1984.)

cortex cells in a vertical "column" to have similarinput resembles the columnar organization of sen-sory cortex cells and is readily understood as aconsequence of the radial distribution of the affer-ent axons.

The input from peripheral receptors is distrib-uted in motor cortex in a somatotopic map, whichis in register with the motor output map shownin Figure 28-4. This map summarizes the sourceof somatic input to most neurons at each corticalsite; however, cells with inputs from other regionsand modalities are considerably intermingled inmotor cortex—more so than in sensory cortex.This heterogeneity of inputs has allowed investi-gators to emphasize different features of the dis-tribution of cells with sensory inputs in motorcortex. For example, in awake macaques, Murphyand colleagues 22 found that neurons driven bystimulation of different forearm regions are dis-tributed in a set of roughly concentric rings (Fig.28-11). In the central region most cells respond toinput from the fingers; surrounding rings containcells responsive to wrist, elbow, and shoulder.These groupings are by no means exclusive, sinceneurons receiving input from adjacent joints weresubstantially intermingled.

In monkeys, the inputs from cutaneous anddeep receptors project to two separate regions ofarea 4. Neurons in a rostral zone of precentralcortex respond predominantly to passive jointmovements, while cells in a more caudal zonerespond primarily to cutaneous input. 32' 33 Thissegregation of proprioceptive (p) cells rostrally andcutaneous (c) cells caudally is illustrated in theschematic sagittal section through pre- and post-central cortex in Figure 28-10. This figure alsoindicates the predominant somatic modality rep-resented in each Brodmann's area. Primary motorand sensory cortical areas are composed of alter-nating zones with predominantly proprioceptiveand cutaneous input, and are bounded rostrallyand caudally by regions that receive visual input(v).

Input-Output Relations

Woolsey's finding37 that the somatotopic map ofperipheral input to motor cortex was in broadregister with the map of output effects evoked bycortical stimulation suggested a close functionalrelationship between input and output, which hasbeen confirmed on the level of single neurons.

Using extracellular microelectrodes Rosen andAsanuma 29 compared the sensory responses ofindividual cortical neurons in the precentral handarea of the squirrel monkey with the movementsevoked by microstimulation at the same sites.Figure 28-12 summarizes their results for a seriesof penetrations in the thumb area. Microstimula-tion evoked different thumb movements at differ-ent sites; the low-threshold sites for evoking flex-ion were distributed in an output zone orientedperpendicular to the cortical surface. Similar zonescould be demonstrated for extension, adduction,and abduction. The sensory responses of neuronsrecorded in each zone were closely related to theoutput; for example, cutaneous receptive fieldswere located on the part of the thumb lying in the

direction of the movement evoked by the stimu-lation. Similarly, cells driven by passive thumbextension were found in the zone whose stimula-tion evoked extension. Similar close relations be-tween the sensory responses of cells and themovement evoked by microstimulation were alsoobserved for the concentric regions illustrated inFigure 28-11.

These input-output relations have a simple func-tional interpretation. The cutaneous input fromreceptors that would be directly activated bymovement could function to assist in grasp re-flexes and placing reflexes by increasing activityof the relevant output cells; indeed, the motorcortex plays an important role in such reflexes.Similarly, the proprioceptive input from stretch

Figure 28-11 Distribution of sensory input from different forelimb joints in motor cortex of the macaque. The precentral gyruswas unfolded onto a flat plane, and boundaries between Brodmann's areas represented by vertical dashed lines. M =medial; L = lateral. Many points also represent output effects on the same joints evoked by microstimulation. (From Murphy,J. T.; Kwan, H. C.; MacKay, W. A.; Wong, Y. C. J. Neurophysiol 41:1120-1131, 1978.)

Figure 28-12 Relation between sensory input and motoroutput evoked by intracortical microstimulation in precentralcortex of monkey. Repetitive stimulation at 5 [La evoked thumbmovement at sites indicated by symbols corresponding tomovements listed at lower right and evoked no movement atsites with bars. Peripheral inputs to cells are indicated bydiagrams representing their cutaneous receptive fields, or bythe description of adequate proprioceptive stimulus; somecells were undriven (UD). (From Rosen, I.; Asanuma, H. Exp.Brain Res. 14:257-273, 1972.)

receptors to cells that facilitate muscle activitycould function to overcome load perturbations ina manner similar to the segmental stretch reflex(see Chap. 24). Further evidence for such a trans-cortical stretch reflex from unit recordings in be-having monkeys is described below.

Coding of Movement Parameters by MotorCortex Cells

While the effects of electrical stimulation andlesions demonstrate that motor cortex plays a rolein controlling movements, these techniques can-not reveal which parameters of movement it con-trols. The movement variables that are coded inthe activity of precentral cortex cells can be inves-tigated only by observing neuronal activity inawake, behaving animals. In such "chronic unitrecording" studies, the activity of single neuronsis recorded with a movable microelectrode. Rele-vant behavioral responses are trained throughoperant conditioning techniques, which gradually"shape" the animal's behavior by selectively re-

warding those responses that are closest to thedesired final behavior. Thus, chronic recordingstudies can document the activity of cells underparticular behavioral conditions designed to testhypotheses about the functions of the recordedneurons. For example, a repeatable movement inresponse to sensory signals is ideal for investigat-ing the relative timing of cell activity involved ingenerating a simple voluntary response. On theother hand, to determine whether motor cortexcells are preferentially involved in coding limbposition or force one can train the animal to movedifferent loads through the same displacement—atask designed to dissociate the force and displace-ment. Experiments can also be designed to dem-onstrate the neural changes involved in preparingto make a particular movement.

Relative Timing of Cell Activity: ReactionTime Responses

One strategy in studying the generation of vol-untary movement has been to determine the rel-ative timing of changes in neural activity in differ-ent motor structures when the animal performs asimple movement. Those structures with neuronsdischarging earlier could presumably "drive"those that are activated later. The timing of neuralactivity is usually investigated by rewarding theanimal for performing a rapid, stereotyped move-ment as soon as possible after detecting a sensorycue. An example of such a reaction-time task is therapid release of a depressed lever when a light isturned on. This paradigm is a voluntary analogueof a segmental reflex, such as the tendon jerk;however, the delay between the stimulus and thebehavioral response—the reaction time—is muchlonger, on the order of 100 to 200 ms. This latencydepends on the amount of training and the stim-ulus modality. Reaction times are increasinglylonger for responses evoked by proprioceptive,auditory, and visual stimuli, respectively. Some ofthis difference is accounted for by the differentafferent conduction times for each sensory modal-ity.

In monkeys trained to release a bar after a visualstimulus, many precentral pyramidal tract cellschange their activity 10 to 100 ms prior to electro-myographic (EMG) activity in the agonist wristmuscles. The changes in activity of these pyrami-dal tract neurons is more tightly linked with initi-ation of muscle activity than with occurrence ofthe light, indicating that they are more related tothe movement than the sensory cue.'°

The routing of neural impulses between a reac-tion time stimulus and the voluntary responseremains largely unresolved. Experiments using thereaction time task have not revealed a simplesequential activation of successive centers. Theproblems are illustrated by experiments designedto determine whether cerebellar cells might pre-cede activation of motor cortex neurons. Figure28-13 shows that the onset times of neural dis-charge in the cerebellar nuclei and motor cortexduring the same motor responses are distributedover several hundred milliseconds and exhibitconsiderable overlap. Similar overlapping distri-butions characterize the onset of activity of cellsin other motor regions. Thus, serial activation ofdifferent motor centers cannot be established bymeasuring neural onset times, because the cellswithin each region—including motoneuronpools—are recruited over times much longer thanthe conduction times between regions. Moreover,the duration of most movements greatly exceeds

Figure 28-13 Relative timing of onsets of change in neuraldischarge in motor cortex, cerebellum, and forelimb musclesduring active wrist movements in a monkey. Although thedistributions show differences in timing, they overlap exten-sively. (From Thach, W. T. J. Neurophysiol. 41:654-674, 1978.)

the conduction times, allowing any recurrent loopsbetween regions to be traversed repeatedly duringa single movement. Consequently, the appealingnotion that initiation of movement involves se-quential activation of cells in hierarchically relatedcenters has proved difficult to confirm experimen-tally. Indeed, the distributions illustrated in Figure28-13 suggest that cells in different regions arerecruited more or less in parallel rather than in astrictly serial manner.

Relation to Active Force

The activity of motor cortex cells during move-ments could potentially code a variety of move-ment parameters. To determine whether activityof pyramidal tract neurons is more strongly relatedto limb position or to the force required to movethe limb, Evarts10 trained monkeys to move ahandle through the same displacement while lift-ing different loads. The discharge of most motorcortex cells was more closely related to the forceexerted or to changes in force than to displacementof the wrist. Similarly, when the monkey wasrequired to hold the handle steady against differ-ent externally applied forces, pyramidal tract neu-rons again discharged in proportion to the iso-metric force exerted.

The fact that activity of motor cortex neuronscovaries with force is still not sufficient evidenceto prove that they are causally involved in gener-ating force. Such a causal linkage can be demon-strated by showing that some of the pyramidaltract neurons also have excitatory effects on ago-nist muscles." Such effects have been demon-strated by the technique of spike-triggered aver-aging, in which the action potentials of a corticalneuron are used to trigger a computer that aver-ages the muscle (EMG) activity occurring in thetime interval after the spike. Some pyramidal tractneurons produce a postspike increase in averageactivity of co-activated limb muscles, as shown inFigure 28-14C. The magnitude and latency of this"postspike facilitation" suggest that this neuron isa corticomotoneuronal (CM) cell and that the ex-tensor carpi ulnaris (ECU) is one of its targetmuscles. Typically, the discharge of CM cells fa-cilitates several co-activated muscles, indicatingthat they exert a divergent influence on a "musclefield."

The discharge of a CM cell facilitates the activityof its target muscles in proportion to its firing rate.Figure 28-14 illustrates the activity of a typical CMcell during wrist movements against an elasticload (which requires a force proportional to dis-

placement). The average of many extension re-sponses (Fig. 28-14B) indicates that this CM cellfires with a phasic burst of activity at movementonset, when extra force is required to overcomeinertial and elastic loads. The initial phasic dis-charge of CM cells begins about 80 ms beforeactivity of their target muscles. This lead time isconsiderably longer than the latency of the post-spike facilitation, which peaks about 10 ms afterthe spike; therefore, much of this early dischargehelps bring the motoneurons to firing threshold.The initial burst is followed by a tonic dischargeduring the period that the monkey exerts a con-stant force. The tonic firing rate of CM cells in-creases with the level of maintained force.' Besidesfacilitating their target muscles, the discharge ofsome CM cells also exerts inhibitory effects onantagonists of the target muscles.

While motor cortex output cells discharge inproportion to active force, this relationship is notinvariant in all behavioral conditions. Pyramidaltract neurons fire more strongly in relation tofinely controlled wrist and finger movements thanto rapid ballistic movements.° Similarly, CM cellsthat facilitate finger muscles are more active duringa precision grip task than during a power griptask, although in the latter condition their target

muscles are even more active?' These observationsare consistent with the behavioral effects of le-sions, which suggest that motor cortex cells areparticularly important in controlling fine fingermovements; these neuronal recordings indicatefurther that pyramidal tract neurons are much lessinvolved in rapid forceful movements.

Load Compensation Responses

The accurate control of voluntary movementinvolves a continual balance between centrallyoriginating command signals and sensory feed-back from the periphery. When a movement issuddenly resisted by an unexpected increase inthe load, several neural mechanisms are activatedto increase the motor output and overcome theperturbation. The stretch of an active muscle by asudden load increase produces a series of muscleresponses. The initial electromyographic response,called Ml, is mediated by the segmental stretchreflex and has a latency of 25 to 30 ms in man(Fig. 28-15). Between 50 and 90 ms there oftenappears a second response, M2, whose mediationhas generated considerable controversy and ex-perimentation. Phillips suggested that this long-latency EMG response could be mediated by CM

Figure 28-14 Response of motor cortex output cell during alternating flexion and extension of wrist. A, From top, activity ofcortical unit and coactivated extensor muscles, wrist torque, and position. 8, Averages of activity and movement parametersfor 100 extension responses. C, Spike-triggered averages of rectified electromyographic (EMG) activity show postspikefacilitation of extensor carpi ulnaris (ECU). (From Fetz, E. E.; Cheney, P. D. J. Neurophysiol. 44:751-774, 1980.)

Figure 28-16 Responses ofmotor cortex cell and elbowmuscles in a monkey during nor-mal elbow flexion (A) and withbrief increases (B) and de-creases (C) of the flexion load.(From Conrad, B.; Matsunami, K.;Meyer-Loman, J.; Wiesendanger,M.; Brooks, V. B. Brain Res. 71:507-514, 1974.)

cells that form part of a transcortical stretch reflex,analogous to the segmental reflex. 25b Indeed, inmonkeys generating active limb movements, un-expected load increases that stretch the agonistmuscle evoke excitatory responses in task-relatedpyramidal tract neurons. For example, the pyram-idal tract neuron in Figure 28-16 fired with activeelbow flexion (A) and responded with an addi-

tional burst when a transient load increase op-posed flexion (B); moreover, the cell's activitydropped following a load decrease, which assistedthe flexion (C). 8 Such responses in CM cells wouldproduce changes in muscle activity that wouldhelp to overcome the load change. While motorcortex output clearly contributes to the "functionalstretch reflex," other pathways and mechanismsmay also be involved. Finally, subjects preparedto resist the perturbation often show a still laterresponse (M3), which is associated with a volun-tary muscle contraction (Fig. 28-15).

Preparatory Set

Prior to execution of a voluntary movement,cortical cell activity may also change in preparationto make the response. For example, a driver wait-ing at a red light is prepared (or set) to step onthe gas pedal when the light changes. Experimentsdesigned to investigate the neural mechanismsinvolved in such a "motor set" have typicallyinvolved monkeys trained to respond after twosuccessive stimuli: the first is an instructional stim-ulus which indicates the type of movement to be

made to a subsequent trigger stimulus. For ex-ample, Evarts and Tanji" used a red light toinstruct the monkey to pull a handle when thehandle was subsequently perturbed, and a greenlight to instruct the monkey to push the handle.During the long delay between the visual instruc-tion stimulus and the proprioceptive trigger stim-ulus caused by the handle perturbation, manymotor cortex cells changed their firing rates, sug-gesting that they were involved in preparing tomake the correct movement. Moreover, the re-sponse of many pyramidal tract neurons to theperturbation of the handle depended on whichmovement the monkey was prepared to make,suggesting that the peripheral input to these cellswas also influenced by the set. The changes in theproprioceptive responses to the perturbation wereusually those that would assist in the subsequentmovement.

Coding of Movement Parameters byPopulation Responses

While most neuronal recording experiments todate have concerned the responses of single neu-rons in relation to movement parameters, thegeneration of a movement clearly involves largepopulations of cells. Thus the coding of responseparameters must involve the coordinated activityof neuronal populations. Recording simultane-ously from a group of motor cortex neurons inmonkeys making wrist movements, Humphreyand colleagues 15 found that an appropriatelyweighted average of their firing rates could matchthe time course of force, and the population av-erage matched force better than the discharge ofany single neuron. Moreover, by using differentsets of weighting factors, the discharges of thesame cells could be made to match also the posi-tion trajectory, or the rates of change of positionor force. The weighting factors derived from oneset of movements would predict the time courseof these parameters in subsequent movements.The match between the predicted and the ob-served movement parameters improved with thenumber of task-related motor cortex cells includedin the weighted average. Figure 28-17C illustratesthe firing rates of five motor cortex cells duringone cycle of wrist movement. Figure 28-17Dshows that the fit between the weighted unitactivity and the time course of the force improvesas more neurons are included in the weightedaverage (bottom to top). This happens because eachadditional cell is added with a weight that wouldfurther decrease the remaining error. More re-

markably, with different weighting factors, theactivity of the same population could be used tomatch a variety of different movement parameters.

The discharge patterns of neuronal populationshave also been used to match the direction of limbmovement in space by a kind of vector addition.Georgopoulos and colleagues" trained monkeysto move a handle from the center of a circle to anyof 12 equidistant points on its circumference. Theyfound that many motor cortex neurons fired withmovements in several directions, but that neuronsusually showed the greatest average activity withmovement in one "preferred" direction. They as-signed a vector to each neuron that pointed in itspreferred direction. Assuming that for every di-rection of movement, the neuron made a vectorcontribution (in its preferred direction) that wasproportional to its mean firing rate during thatmovement, they found that the sum of the vectorsof the neuronal population pointed in the directionof the limb movements. Again, the direction pre-dicted by the vector sum of the population dis-charge became more accurate as more cells wereincluded.

These studies indicate that close correlates ofvarious movement parameters may be derivedfrom the activity of neuronal populations. More-over, the appropriate function of the populationactivity matches the mechanical parameters withincreasing accuracy as the number of cells includedis increased. Although such functions of popula-tion discharge can provide good descriptivematches with movement variables, this does notprovide a causal explanation of how the nervoussystem controls these parameters.

Effects of Motor Cortex Lesions

Spasticity and Paralysis

Lesions of motor cortex in humans may resultfrom cortical strokes, missile wounds, or surgicalexcisions required to treat epilepsy or to removetumors. Damage to primary motor cortex initiallyproduces flaccid paralysis of the body parts rep-resented in the affected region. Within one to twoweeks the ability to move the proximal joints isoften regained. At the same time spasticity ap-pears, becoming most severe and permanent inthe distal muscles; spastic muscles are hyperactiveand show exaggerated stretch reflexes. The mostsevere and long-lasting deficits occur in the exten-sors of the wrist and fingers.

1 sec

Figure 28-17 Population coding of movement parameters. A, Monkey's task involved alternating wrist movements against aconstant load. B, Firing rate of cortical neurons was smoothed by a filter as shown. C, Smoothed firing rates of five simultaneouslyrecorded motor cortex cells during a single trial (bars = 10 spikes/sec); the antidromic latencies of the four pyramidal tractneurons are given at left. Bottom traces show force (solid line) and displacement (dashed line). D, Matches between the forcetrace and the weighted averages of different numbers of cells: Bottom, single cell (No. 5); middle, three cells (Nos. 3-5); top;all five cells (Nos. 1-5). (From Humphrey, D. R.; Schmidt, E. M.; Thompson, W. D. Science 170:758-762. Copyright 1970 by theWs.)

Damage to primary motor cortex also impairsthe ability to move joints independently. For ex-ample, after a lesion in the foot area, the ankletypically cannot be extended without concomitantflexion of the knee. Also, after damage to theprecentral hand area, the fingers can be flexedonly as a group. These muscle "synergies" per-manently replace the normal independent controlof joints.

In contrast with clinical cases, which often in-volve additional complications, experimental abla-tion or cooling studies in animals provide morecontrolled conditions for testing cortical function. 9

The effects of cortical ablation are less severe incarnivores than in primates. After total decortica-tion, a cat or dog can regain the ability to standand walk, although it permanently loses tactileplacing reflexes (the ability to correctly place the

paw on a surface following tactile contact) andhopping reactions (the ability to reposition thepaw when moving relative to the surface). Incarnivores, subcortical centers are sufficient togenerate righting responses and locomotion, butthe cortex is necessary to perform learned re-sponses appropriate to the external environment.

Primates are more severely and permanentlyaffected by cortical damage. Ablation of the pre-central hand area in the chimpanzee initially pro-duces complete paralysis of the hand. After amonth, crude grasping responses return, but themovements remain slow and inept and the fingersshow persistent spastic flexion. Removal of theentire motor cortex produces maximal paralysisand spasticity, with hypertonia in flexors of thearm and extensors of the leg. In higher primates(apes and humans), lesions of motor cortex also

produce a positive Babinski reaction. While tactilestimulation of the foot pad normally evokes plan-tar (downward) flexion of the toes, after lesions ofprecentral cortex or the pyramidal tract the samestimulus evokes upward movement of the footand toes. Babinski's sign has become a classicclinical test for damage to the pyramidal system.

Recovery from cortical ablation depends uponseveral factors. In primates the consequences ofcortical removal can be considerably less severe ifthe cortex is ablated in stages which are separatedby months or years. Following such gradual de-cortication, some monkeys recover the ability tomake rudimentary movements. Also, recoveryfrom cortical injury is considerably greater whenthe damage occurs early in life. An infant monkeydeprived of its precentral cortex compensates al-most completely for the loss.

Effects of Pyramidal Tract Lesions

The corticospinal tract may be selectively andcompletely sectioned at the medullary pyramids,producing an animal whose remaining motor func-tions are subserved by the extrapyramidal system.The behavioral consequences of pyramidal tractsection were extensively described by Tower: 2

The most conspicuous result of unilateral pyramidallesion in the monkey is diminished general usage andloss of initiative in the opposite extremities. . . . Whenboth sides are free to act, initiative of almost every sortis delegated to the normal side, but if the normal sideis restrained, the affected side can, with sufficientlystrong excitation, be brought to act.

The affected side shows a paresis—a loss of finecontrol of movement, particularly in the distalmuscles. The skillful use of the hand in precisemovements is entirely lost. Although the limb canstill be used in postural activity and in reachingand grasping, the movements are clumsy andrequire considerable attention and effort.

Unilateral pyramidal section also produces hy-potonia in muscles of the opposite side. Contra-lateral muscles exhibit abnormally low resistanceto passive stretch and slowed tendon reflexes.Reflexes evoked by cutaneous stimulation alsohave higher thresholds. Reactions to pinprick andabdominal reflexes either are abolished or areharder to elicit and slower. Contact placing isabolished and proprioceptive placing is more dif-ficult to elicit. The ability to reach for and graspobjects remains but is much clumsier. Also, theability to manipulate objects and release a graspis impaired.

Pyramidal lesions in the chimpanzee producesimilar but more severe deficits. Discrete controlof the digits is profoundly impaired, and the graspreflex becomes so hyperactive that it is impossibleto release objects. The chimpanzee, like humans,shows the classic Babinski sign after pyramidalsection.

MOTOR FUNCTIONS OF OTHER CORTICALAREAS

A goal-directed movement, such as reaching foran object, requires that the activities of manymuscles be coordinated in a manner appropriateto environmental conditions. While primary motorcortex is involved mainly in executing movements,other cortical areas are more involved in the priorstages of programming the limb movements to beappropriate for the particular context or goal. Pro-gramming the patterns of muscle activity is largelyaccomplished by three interconnected corticalareas: the supplementary motor cortex, the pre-motor cortex, and the posterior parietal cortex (seeFigs. 28-1 and 28-9). Each region appears to spe-cialize in different aspects of movement contro1.36

Supplementary Motor Cortex

The supplementary motor area (SMA) lies an-terior and medial to primary motor cortex, largelyin the medial surface of the hemisphere. Move-ments evoked by electrical stimulation of the SMArequire higher intensities and longer stimulustrains than those evoked from precentral cortex.Like MI, the SMA is somatotopically organized,with the head anterior and hindlimb posterior.The evoked movements are often more complexand prolonged than the simple muscle contrac-tions obtained from precentral cortex. For exam-ple, evoked responses may involve orientation ofthe limb or body, or coordinated movements suchas opening the hand. These movements may out-last the duration of the stimulus and sometimesare elicited on both sides of the body. Neurosur-geons stimulating the SMA in awake patients haveevoked vocalization with associated facial move-ments, coordinated movements of the limbs, andalso inhibition of voluntary movements.

Excision of the SMA in human patients hasresulted in transient speech deficits, or aphasias,which typically disappear after several weeks. Theloss of the SMA also results in a persistent slow-

ness in generating repetitive movements. Lesionsof the SMA in monkeys has produced interestingapraxias of reaching and bimanual coordination.Monkeys with unilateral ablation of the SMA can-not coordinate their hands in a bimanual task.Faced with a horizontal plastic plate containingholes stuffed with raisins, a normal monkeyquickly presses the raisins out with one hand andcatches them with the other hand cupped below.After a unilateral lesion of the SMA the two handscannot be coordinated independently, but insteadmove together in a similar manner, as if the intactSMA now controlled both hands. Sectioning thecorpus callosum abolishes the bimanual deficit,suggesting that each SMA normally communicateswith both hemispheres. 3

Combined unilateral lesions of the SMA andpremotor cortex abolish the ability of a monkey toreach around a transparent barrier to obtain avisible slice of apple with its contralateral hand. 19

Instead, the monkey persists in reaching for theapple directly and repeatedly hits the plastic plate.The same monkey has no problems performingthe task using its other hand, indicating that itscomprehension and motivation are intact. Thisapraxia lasts for at least two years. In contrast, amonkey with a lesion of primary motor cortex canreach around such a barrier, albeit clumsily.

Single unit recordings in the SMA of consciousmonkeys also suggest that SMA cells may play arole in coordinating movements of the two limbs.Many cells fire in a similar way for comparablemovements of the ipsilateral and contralateral arm.Such bilateral responses are found for cells relatedto distal as well as proximal joint movements.Some SMA cells respond to somatic stimulation;these are often driven by manipulation of multiplejoints or bilateral somatic stimuli.

That the SMA is involved in programming se-quences of movements has been revealed by meas-uring cerebral blood flow in conscious humansubjects. Increases of cortical neural activity iscorrelated with localized increases in blood flow,which can be detected by monitoring the circula-tion of radioactive xenon with an array of radiationdetectors around the scalp." Subjects performingthe simple motor task of squeezing a spring be-tween the thumb and forefinger showed clearincreases in blood flow in both the primary motorand sensory cortex contralateral to the active hand.When they performed a more complex sequenceof finger movements, touching the thumb succes-sively with each of the other fingers in turn, theirblood flow increased over the SMA bilaterally aswell as over sensorimotor cortex. Furthermore,

when the subjects simply thought about the se-quential movements without performing them,regional blood flow again increased over the SMA,but not over sensorimotor cortex, suggesting thatthe SMA is involved in programming complexsequential movements.

Premotor Cortex

The premotor cortex (PMC) lies anterior to pri-mary motor cortex on the lateral surface of thehemisphere in area 6 (see Fig. 28-1). PMC iscomparable in size to MI in monkeys, but inhumans it is almost six times larger than MI.Unlike MI and the SMA, PMC makes only a minorcontribution to the corticospinal tract; instead, itsdescending output is directed largely to the med-ullary reticular formation. Electrical stimulation ofPMC is much less likely to evoke movements thanstimulation of MI, or even of the SMA. The elicitedmovements often involve the proximal muscula-ture and require much higher stimulus intensitiesthan those evoked from MI or the SMA.

In humans, lesions involving PMC produceweakness in shoulder and hip muscles and diffi-culty in abduction and elevation of the contralat-eral arm. Limb movements are also slower, asevidenced by delayed muscle activation. PMC le-sions may also produce an inability to move thetwo arms simultaneously in a coordinated fashion(called movement-kinetic apraxia).

Like some cells in MI, many neurons in PMCdischarge when a monkey is preparing to make aparticular movement. In the experiment illustratedin Figure 28-18, 35 a monkey was trained to depressthe one key of a set that was illuminated (theinstructional stimulus), but only after anothersmall light (the trigger stimulus) had been turnedon. Most of the task-related cells in PMC, likethose in MI, responded during the execution ofthe movement; others responded to the instruc-tional cues (see also Fig. 31-13). However, certainPMC cells exhibited a sustained increase in dis-charge throughout the delay between the instruc-tional cue and the trigger stimulus. This set-relateddischarge was often directionally specific, occur-ring only when the monkey was preparing tomove in a particular direction. This discharge wasnot simply related to a subliminal excitation ofagonist motoneurons, since it typically stoppedwith the onset of the movement. PMC contains ahigher proportion of cells related to motor set thandoes MI. Similar "set-related" cells also have beenobserved in the SMA and prefrontal cortex.

Figure 28-18 Set-related activ-ity in a premotor cortex cell. Top:Behavioral paradigm. The mon-key rested its hand on one offour keys during the intertrial pe-riod (1), observed the ready lightappear at another key (2), thenat a random time saw a go sig-nal (3) and moved to the illumi-nated key (4). Bottom: Set-re-lated discharge in this premotorcell occurred whenever themonkey was prepared to moveto the right, whether the cue wasvisual (left) or auditory (right). Thiscell was also inhibited after in-structions to move to the left(bottom). (From Weinrich, M.;Wise, S. P. J. Neurosci. 2:1329-1345, 1982.)

Frontal Eye Fields

The frontal eye fields lie anterior to the PMC inBrodmann's area 8 (see Fig. 28-1). Efferent projec-tions from the frontal eye fields travel via theinternal capsule to pontine regions controlling eyemovements and to the superior colliculus, whichalso is involved in the generation of saccadic eyemovements (see also Fig. 20-19).

Electrical stimulation of the frontal eye fields inprimates and carnivores typically causes conjugatemovement of both eyes to the opposite side; stim-ulation at some sites may cause the eyes to moveobliquely upward or downward, with the horizon-tal component away from the stimulated side.Stimulation of a particular site in the frontal eyefields evokes saccades with a characteristic direc-tion and amplitude, which are largely independentof the stimulus parameters and the initial positionof the eyes in the orbit. Thresholds for evokingsaccades are raised if the subject is fixating ortracking a visual target. Stimulation of the frontaleye fields may also evoke head movement to theopposite side and can sometimes produce auto-nomic responses such as dilatation of both pupilsand lacrimation. 2

Unilateral ablation of area 8 produces a sus-tained deviation of both eyes toward the side ofthe lesion and impairs voluntary eye movementsto the opposite side. Unilateral lesions of area 8may also produce deviation of the head to the sideof the lesion. Bilateral ablation of the frontal eyefields can impair the ability to gaze laterally ineither direction. Although the eyes are capable offollowing a moving target, they always drift backto the central position. Bilateral ablation can alsodestroy the ability to attend visually to objects inthe peripheral visual field. These deficits are usu-ally temporary, and eye movements often recoverwithin several days after the lesion.

The activity of neurons in the frontal eye fieldsis related to eye or head movements. Many cellsfire before or after voluntary saccades in a givendirection. 4 Cells which discharge before voluntarysaccades in a particular direction have been re-corded at the sites with lowest thresholds forevoking those saccades. While many cells exhibita burst of firing immediately before the saccade,certain "visuomovement" cells exhibit sustainedactivity between presentation of a visual targetand the saccade to the target, reminiscent of thebehavior of certain PMC neurons. Other frontaleye field neurons discharge with gaze in a certaindirection, as well as pursuit in this direction. Athird type of cell in Brodmann's area 8 is relatedexclusively to head movements.

Taken together, this evidence suggests that thefrontal eye fields are involved in generating vol-untary gaze and saccades. However, their role isnot completely analogous to that of area 4 for limbmovements, since lesions of the frontal eye fieldscause only temporary impairment of eye move-ments. The frontal eye field cells are probablyinvolved in mediating visually evoked saccades,but they apparently exert their influence in con-junction with or through the superior colliculus(see also Chap. 20).

Posterior Parietal Cortex

In addition to the anterior premotor regionsdescribed above, the posterior parietal cortex(areas 5 and 7) also appears to be involved inprogramming limb movements. Humans with le-sions of posterior parietal cortex show classic signsof an apraxia—an inability to make directed limbmovements in a particular context, in the absenceof motor weakness. Their symptoms can be de-scribed as a sensory and motor neglect of theopposite hemifield of extrapersonal space. Theycannot reach accurately for objects and appear toneglect information from the contralateral hemi-field. Such neglect is particularly pronounced withparietal lesions of the nondominant hemisphere(see Chap. 31).

Mountcastle and colleagues 20 have shown thatthe activity of certain neurons in areas 5 and 7 ofmonkeys are specifically related to active reachingmovements. Some neurons discharged only whenthe monkey reached for a desirable object, suchas food, in its immediate extrapersonal space, butdid not fire during similar limb movements thatwere not directed toward acquiring an object ofinterest. A related class of cells fired preferentiallyduring active manipulation of objects of interest.Area 7 contains comparable oculomotor cells thatfire specifically when the monkey moves its eyestoward an object of interest, but not during spon-taneous eye movements. These workers suggestedthat these posterior parietal neurons may generatea motor command to acquire objects of interest,either manually or visually, in extrapersonal space.(An alternative interpretation is described in thediscussion of the parietal lobe in Chap. 31).

DISTRIBUTED CORTICAL MOTOR FUNCTION

This chapter has adopted the view that primarymotor cortex is preferentially involved in the exe-cution of movements, whereas secondary motor

3.

3a.

4.

5.

6.

7.

8.

12.

13.

19.

areas are more involved in programming. While 2.this dichotomy provides a useful framework, theview that these functions are exclusively localizedin separate cortical regions must be recognized asoverly simplistic. Some of the evidence discussedsuggests that generation of movement involvescontinual interactions between- different corticalregions. First, the neural connections betweenprimary and secondary cortical areas are recipro-cal, not serial, and each area has its own subcor-tical input and output connections. Second, al-though the various motor regions have neuronswith response properties consistent with their pro-posed function, each area actually contains a mix-ture of cell types, all of which are present to someextent in the other areas. Cortical neurons withsimilar response properties might be intercon-nected and act together. Such a distributed rep-resentation could explain the recovery of somefunctions that are temporarily lost after lesions.Finally, the relative timing of cell responses indifferent regions during a reaction time movementlargely overlap, suggesting that all areas are acti-vated more or less simultaneously. Thus, attrib-uting the functions of motor programming andexecution to different cortical regions is more amatter of relative degree than an absolute dichot-omy. In a similar way, cortical and subcorticalcenters must also interact continuously and coop-eratively in producing movements.

14.

BIBLIOGRAPHY

Recommended Reviews

Evarts, E.V. Role of motor cortex in voluntary movement.Handbook of Physiology. 2:1083-1121, 1981.Review of motor cortex cell activity in relation to movements andtranscortical reflexes.

Humphrey, D.R. Cortical control of reaching. In Talbott, R.E.;Humphrey, D.R., eds. Posture and Movement, 51-112. NewYork, Raven Press, 1979.The roles of secondary motor cortex areas in programming limbmovements.

Phillips, C.G.; Porter, R. Corticospinal neurones. Their role inmovement. Monographs of the Physiological Society, Lon-don, Academic Press, 1977.A comprehensive review of the role of motor cortex and pyramidaltract in control of movement.

20.

REFERENCES21.

1. Asanuma, H.; Rosen, I. Topographical organization ofcortical efferent zones projecting to different forelimb mus- 22.cles in the monkey. Exp. Brain Res. 14:243-256, 1972.

Bucy, P.C., ed. The Precentral Motor Cortex. Urbana, Ill.,Univ. of Illinois Press, 1949.Brinkman, C. Supplementary motor area of the monkey'scerebral cortex: Short- and long-term deficits after unilateralablation and the effects of subsequent callosal section. J.Neurosci. 4:918-929, 1984.Brodmann, K. Vergleichende Lokalisationslehre der

Grosshirnrinde. Leipzig, J.A. Barth, 1909.Bruce, C.J.; Goldberg, M.E. Physiology of the frontal eyefields. Trends Neurosci. 7:436 441, 1984.Chang, H.T.; Ruch, T.C.; Ward, A.A., Jr. Topographicrepresentation of muscles in motor cortex in monkeys. J.Neurophysiol. 10:39-56, 1947.Cheney, P.D.; Fetz, E.E. Functional classes of primatecorticomotoneuronal cells and their relation to active force.I. Neurophysiol. 44:775-791, 1980.Clough, J.F.M.; Kernell, D.; Phillips, C.G. The distributionof monosynaptic excitation from the pyramidal tract andfrom primary spindle afferents to motoneurones of thebaboon's hand and forearm. J. Physiol. (Lond.) 198:145-166,1968.Conrad, B.; Matsunami, K.; Meyer-Loman, J.; Wiesendan-ger, M.; Brooks, V.B. Cortical load compensation duringvoluntary elbow movements. Brain Res. 71:507-514, 1974.Denny-Brown, D. The Cerebral Control of Movement. Spring-field, Ill., Charles C Thomas, 1966.Evarts, E.V. Relation of pyramidal tract activity to forceexerted during voluntary movement. J. Neurophysiol. 31:14-27, 1968.Evarts, E.V.; Tanji, J. Reflex and intended responses inmotor cortex pyramidal tract neurons of monkey. J. Neu-rophysiol. 39:1069-1080, 1976.Fetz, E.E.; Cheney, P.D. Postspike facilitation of forelimbmuscle activity by primate corticomotoneuronal cells. J.Neurophysiol. 44:751-772, 1980.Fromm, C.; Evarts, E.V. Relation of motor cortex neuronsto precisely controlled and ballistic movements. Neurosci.Lett. 5:259-266, 1977.Georgopoulos, A.P.; Kalaska, J.F.; Crutcher, M.D.; Cam-initi, R.; Massey, J.T. The representation of movementdirection in the motor cortex: single cell and populationstudies. In Edelman, G.; Gall, W.E.; Cowan, W.M., eds.Dynamic Aspects of Neocortical Function, 501-529. New York,John Wiley & Sons, 1984.Humphrey, D.R.; Schmidt, E.M.; Thompson, W.D. Pre-dicting measures of motor performance from multiple spiketrains. Science 170:758-762, 1970.Jankowska, E.; Padel, Y.; Tanaka, R. Projections of pyram-idal tract cells to a-motoneurones innervating hind-limbmuscles in the monkey. J. Physiol. 249:637-667, 1975.Jones, E.G. Connectivity of the primate sensory-motorcortex. In Jones, E.G.; Peters, A., eds. Cerebral Cortex, vol.5, 113-184. New York, Plenum Press, 1986.Jones, E.G.; Wise, S.P. Size, laminar and columnar distri-bution of efferent cells in the sensorimotor cortex of mon-keys. J. Comp. Neurol. 175:391-438, 1977.Moll, L.; Kuypers, H.G.J.M. Premotor cortical ablations inmonkeys: contralateral changes in visually guided reachingbehavior. Science 198:317-319, 1977.Mountcastle, V.B.; Lynch, J.C.; Georgopoulos, A.; Sakata,H.; Acuna, C. Posterior parietal association cortex of themonkey: command functions for operations within extra-personal space. J. Neurophysiol. 38:871-908, 1975.Muir, R.B.; Lemon, R. N. Corticospinal neurons with aspecial role in precision grip. Brain Res. 261:312-316, 1983.Murphy, J.T.; Kwan, H.C.; MacKay, W.A.; Wong, Y.C.Spatial organization of precentral cortex in awake primates.

III. Input-output coupling. J. Neurophysiol. 41:1120-1131,1978.Olszewski, J. The Thalamus of the Macaca Mulatta: An Atlasfor Use with the Stereotaxic Instrument. Basel, Karger, 1952.

4. Patton, H.D.; Amassian, V.E. The pyramidal tract: itsexcitation and functions. Handbk. Physiol. 2:837-861, 1960.

5. Penfield, W.; Rasmussen, A.T. Cerebral Cortex of Man. AClinical Study of Localization of Function. New York, Mac-millan, 1950.

5a. Phillips, C.G. Motor apparatus of the baboon's hand.Proc. R. Soc. Lond. [Biol.] 173:141-174, 1969.

5b. Phillips, C.G.; Porter, R. Cortical spinal neurones. Theirrole in movement. Monographs of the Physiological Soci-ety. London, Academic Press, 1977.

6. Phillips, C.G.; Porter, R. The pyramidal projection tomotoneurones of some muscle groups of the baboon'sforelimb. Prog. Brain Res. 12:222-242, 1964.

7. Porter, R. The corticomotoneuronal component of thepyramidal tract: corticomotoneuronal connections andfunctions in primates. Brain Res. Rev. 10:1-26, 1985.

8. Roland, P.E.; Larsen, B.; Lassen, N.A.; Skinhoj, E. Sup-plementary motor area and other cortical areas in organi-zation of voluntary movements in man. J. Neurophysiol.24:91-100, 1980.

9. Rosen, I.; Asanuma, H. Peripheral afferent inputs to theforelimb area of the monkey cortex: input-output relations.Exp. Brain Res. 14:257-273, 1972.

3. Schell, G.R.; Strick, P.L. The origin of thalamic inputs tothe arcuate premotor and supplementary motor areas. J.Neurosci. 4:539-560, 1984.

I. Shinoda, Y.; Yokota, J-I.; and Futami, T. Divergent projec-tion of individual corticospinal axons to motoneurons of

multiple muscles in the monkey. Neurosci. Lett. 23:7-12,1981.

31a. Stoney, S.D.; Thompson, W.D.; Asanuma, H. Excitationof pyramidal tract cells by intracortical microstimulation:effective extent of stimulating current. J. Neurophysiol.31:659-669, 1968.

32. Strick, P.L.; Preston, J.B. Two representations of the handin area 4 of a primate. J. Neurophysiol. 48:139-159, 1982.

33. Tanji, J.; Wise, S.P. Submodality distribution in sensori-motor cortex of the unanesthetized monkey. J. Neurophys-iol. 45:467-481, 1981.

34. Thach, W.T. Correlation of neural discharge with patternand force of muscular activity, joint position and directionof intended next movement in motor cortex and cerebel-lum. J. Neurophysiol. 41:654-676, 1978.

35. Weinrich, M.; Wise, S.P. The premotor cortex of themonkey. J. Neurosci. 2:1329-1345, 1982.

36. Wiesendanger, M. Organization of secondary motor areasof the cerebral cortex. Handbk. Physiol. 2:1121-1147, 1981.

36a. Wiesendanger, M. The pyramidal tract: its structure andfunction. In Towe, A.L.; Luschei, E.S., eds. Handbook ofBehavioral Neurophysiology, vol 5, Motor Coordination, 401-491. New York, Plenum Press, 1981.

37. Woolsey, C.N. Organization of somatic sensory and motorareas of the cerebral cortex. In Harlow, H.F.; Woolsey,C.N., eds. Biological and Biochemical Bases of Behavior.Madison, Wis., Univ. of Wisconsin Press, 1958.

38. Woolsey, C.N.; Gorska, T.; Wetzel, A.; Erickson, J.C.;Earls, F.J.; Allman, J.M. Complete unilateral section of thepyramidal tract at the medullary level in Macaca mulatta.Brain Res. 40:119-123, 1972.