Progress in Neurobiology, Vol. 50, pp.381 to 425, 1996 Copyright~ 1996ElsevierScienceLtd. All rightsreserved Printed in Great Britain 0301-0082/96/$32.00

Pergamorr @

PII: S0301-0082(96)00042-1

THE BASAL GANGLIA: FOCUSED SELECTION AND INHIBITION OF COMPETING MOTOR PROGRAMS JONATHAN W. MINK Department

of Neurology,

Box 8111, Washington University School of Medicine, St. Louis, MO 63110, U.S.A. (Received

17 January

1996; Accepted

660 S. Euclid Ave.,

23 June 1996)

Abstract-Thebasalgangliacompriseseveralnucleiintheforebrain,diencephalon, andmidbrainthought to playa significantrolein thecontrolof postureand movement.It is wellrecognizedthat peoplewith degenerativediseasesof the basalgangliasufferfromrigidlyheldabnormalbodypostures,slowingof movement,involuntarymovements,or a combinationof theseabnormalities.However,it has not been agreedjust whatthebasalgangliacontributeto normalmovement.Recentadvancesin knowledge of the basalgangliacircuitry,activityofbasalganglianeuronsduringmovement,andtheeffectof basalganglia lesionshaveledto a newhypothesisofbasalgangliafunction.Thehypothesisstatesthatthebasalganglia do not generatemovements.Instead,whenvoluntarymovementis generatedby cerebralcorticaland eerelwllarmechanisms, thebasalgangliaact broadlyto inhibitcompetingmotormechanisms that would otherwiseinterferewiththe desiredmovement.Simultaneously, inhibitionis removedfocallyfromthe desiredmotor mechanismsto allowthat movementto proceed.Inabilityto inhibitcompetingmotor programsresultsin slowmovements,abnormalposturesand involuntarymuscleactivity.Copyright~ 1996ElsevierScienceLtd. CONTENTS 1. Introduction 2. Anatomyof thebasalganglia 2.1. Excitatoryinputto striatumfromcerebralcortex 2.2. Focusedinhibitoryoutputfromstriatumto basalgangliaoutputnuclei 2.3. Heterogeneous intrinsicorganizationof thestriatum 2.4. Mechanismsthat focusstriatalactivity 2.4.1.Spatialandtemporalfocusingbymediumspinystriatalneurons 2.4.2.Dopamineinputfromsubstantial nigraparscompactamodulatescorticalinputs 2.4.3.Mediumspinyneuronshaveextensivelocalaxoncollaterals 2.4.4.Cholinergicinputregulatesthemembranestateof mediumspinyneurons 2.5. Excitatoryinputfrommotorareasofcerebralcortexto subthalamicnucleus 2.6. Divergentexcitatoryinputfromsubthalamicnucleusto basalgangliaoutputnuclei 2.7. Criticaldifferences betweenthesubthalamicandstriataloutputs 2.8. GIobuspallidusparsinternaandsubstantial nigraparsreticulateareoutputnuclei 2.8.1.Inhibitoryoutputfromglobuspallidusparsinternato thalamusandbrainstem 2.8.2.Inhibitoryoutputfromsubstantial nigraparsreticu[atato thalamus,superiorcolliculus andbrainstem 2.9. Globuspallidusparsexternaandsubstantial nigraparscompacta 2.9.1.Globuspallidusparsexternareceivesinputfromstriatumand STNandsendsoutputto STN, GPiand SNpr 2.9.2.Substantial nigraparscompactais reciprocally connectedwiththestriatum 2.10.Basalgangliaanatomy:summaryandsynthesis 3. Activityof thebaaalgangliaduringmovement 3.1. Neuronalactivityin theputamencorrelateswithlimbmovement 3.1.1.Movement-related striatalneuronsaresomatotopically organized 3.1.2.Movement-related striatalneuronsfireaftermovementhasbeeninitiated 3.1.3.Somestriatalneuronshave“set”signals 3.2. Caudateactivitycorrelatedwitheyemovements 3.3. Context-dependent neuronalactivityin thestriatum 3.4. Striatalactivityrelatedto anticipationof “behaviorally significant”events 3.5. Subthalamicnucleusneuronsaretonicallyactiveandsomatotopically organized 3.5.1.Subthalamicnucleusactivityrelatedto limbmovement 3.5.2.Subthalamicnucleusactivityrelatedto saccades 3.6. Globuspallidusparsinternaand substantial nigraparsretictdataaresomatopically organized 3.6.1.Globuspallidusparsinternaneuronsaretonicallyactiveduringmaintainedpostureand changeafterlimbmovementisinitiated 3.6.2.Globuspallidusinternalsegmentactivitydoesnotcorrelatewithphysicalparametersof movement 3.6.3.GPineuronalactivityis notexclusively relatedto anyonemovementmodeor context 3.6.4.$ubstantianigraparsreticr.data neuronsdischargein relationto eyemovements 381

382 383 383 384 385 386 386 386 388 388 388 388 389 389 389 389 390 390 390 390 390 391 391 392 392 392 393 394 395 395 395 396 396 397 397 399

382

J. W. Mink

CONTENTS (Ccwirrued) 3.7. Dischargeof globuspallidusexternalsegmentand internalsegmentare similar 3.8. Substantianigrapars compactaactivityisrelated to certainbehavioralevents 3.9. Summaryofsignalingin the basalganglia 4. The effectof basalgangliadamage 4.1. Movementdisordersassociatedwithbasalgangliadisease 4.1.1.Parkinson’sdisease 4.1.1.1.Bradykinesia(slowingof movement) 4.1.1.2.Rigidityand posturalinstability:inabilityto inhibitreflexmechanisms 4.1.1.3.Akinesia 4.1.2.Huntington’sdiseaseand chorea 4.1.3.Hemiballismus due to subthalamicnucleusdamage 4.1.4.Dystonia 4.2. Movementabnormalitiesresultingfromexperimentallesionsof the basalganglia 4.2.1.Lesionsof striatumproducevariabledeficits 4.2.1.1.Experimentallesionsof the putamen 4.2.1.2.Experimentallesionsof the caudate 4.2.1.3.Experimentalstriatallesionsrarelyproducechorea 4.2.2.Experiments]lesionsof the subthalamicnucleusproducehemiballismus or chorea 4.2.3.Experimentallesionsof globuspallidus 4.2.3.1.Combinedlesionsof GPeand GPi 4.2.3.2.Experimentallesionsrestrictedto GPi 4.2.3.3.Effectof GPi lesionson the activityof downstreamtargets 4.2.4.Experimentallesionsof substantialnigrapars reticulateimpaireyemovements 4.2.5.Experimentallesionsrestrictedto globuspallidusparsexterna 4.2.6.Experimentallesionsof substantialnigrapars compactamimicParkinson’sdisease 5. Focusedselectionand inhibitionof competingmotorprograms:an hypothesisof basalgangliafunction 5.1. Whymustcompetingmotorpatterngeneratorsbe inhibitedduringa movement 5.2. The hypothesiscan accountfor the syndromesof basalgangliadysfunction 5.2.1.Pallidalablation 5.2.2.Chorea/hemiballismus 5.2.3.Parkinsonism 5.3. The schemeof focusedselectionand inhibitionof competingmotormechanismsand its relationship to otherrecentmodelsof basalgangliafunction 5.3.1.Initiationof movementby the basalganglia 5.3.2.Do the basal~andiascalemovementby ouposing“direct”and “indirect”pathways? 5.3.3.Do the basal~an~liasequencemovement? -6. Conclusion Acknowledgements References

1. INTRODUCTION Participation of the basal ganglia in the control of movementis an old idea dating back to the early part of this century when S.A. Kinnier Wilson described a disease characterized by muscular rigidity, tremor, and weaknesswith pathological changes in the liver and basal ganglia (hepatolenticular degeneration) (Wilson, 1912). Wilson noted that several features usually associated with damage to the pyramidal tracts were not present in this diseaseand postulated that the motor abnormalitiesweredue to dysfunction of an “extrapyramidal” motor system. Wilson postulated that the basal ganglia were the major constituents of this extrapyramidalmotor systemthat he viewed to be parallel to and independent of the pyramidal (corticospinal) motor system. Wilson developed the view of two motor systems, the phylogenetically “old” extrapyramidal system and the “new” pyramidal system (Wilson, 1928). He thought that the function of the extrapyramidal systemwas automatic, postural, static, and minimally modifiable and that the function of the pyramidal systemwas voluntary, phasic, and readilymodifiable. In the 1960s, more modern anatomical techniques showed that the bulk of the basal ganglia output went via thalamus to motor cortical areas

399 401 402 402 402 402 403 403 404 404 405 405 405 405 405 406 406

407 407 407 408 409 409 411 411 414 414 415 415 415 416 416 416 417 418 418 419 419

(Nauta and Mehler, 1966).Thus, it appeared that the basal ganglia output was prepyramidal rather than extrapyramidal. This finding contributed to the development of hypotheses that the basal ganglia initiate movement via their output to motor cortex. More recxmtly, it was shown that the majority of the basal ganglia output goes to both thalamus and brainstem (Parent and De Bellefeuille, 1982)so that the basal ganglia may be both “prepyramidal” and “extrapyramidal”. Even more recent data have indicated that the output of the basal ganglia goes to many areas of the frontal lobes, including areas that are generally thought to have cognitive functions (Alexander et al., 1986; Middleton and Strick, 1994).However, the role of the basal ganglia in motor control appears to be the dominant one, and it will be the emphasis of this review. When voluntary limb movement is studied, attention is usually focused on mechanismsinvolved in the planning and execution of that particular movement. However,during any given movement, a multitude of potentially competing motor mechanisms must also be inhibited to prevent them from interfering from the desired movement. Consider as an examplethe act of reachingto pick an apple from a tree. If asked to describe the movement, most

TheBasalGanglia observers would describe the reaching movement. Yet, during the reach, multiple other motor mechanisms act together to maintain the upright posture of the rest of the body. Prior to the reach, these mechanisms were also active in the reaching arm to maintain its posture. When the arm reaches toward the apple the postural mechanisms must be turned off selectivelyin the arm while they remain active in the rest of the body. When the reach is completed, the reaching motor pattern generators (MPGs) must be turned off and the postural mechanisms must be turned back on. If the competingposture-holdingand reachingmechanisms were inappropriately active at the same time, the result might be instability of posture or slowing of movement, or both. Thus, when a movement is generated, it is not sufficient to just turn on the desired MPGs. MPGs which had previously been active or that might otherwise compete with the intended movement must also be turned off. In this review,it will be proposed that the inhibitory output of the basal ganglia acts selectively to inhibit competing motor mechanisms in order to prevent them from interferingwith voluntary movementsthat are generated by other central nervous system structures. This review is organized into 4 sections. The first is an overview of recent progress in basal ganglia anatomy. The second summarizes the activity of single neurons in the basal ganglia during limb and eye movement. The third describes the movement abnormalities associated with human basal ganglia disease and reviewsthe results of experimentalbasal ganglia lesions in non-human primates. In the final section a scheme of focal selectionand inhibition of competing motor mechanisms is developed.

383

striatum and sends the bulk of its output back to striatum. 2.1. Excitatory Input to Striatum from Cerebral Cortex The striatum receivesan excitatory, glutamatergic input from all of cerebral cortex except for primary visual and auditory areas (Cherubini et al., 1988; Kemp and Powell, 1970;Kitai et al., 1976;McGeer et al., 1977).The cortical input is organized with a rough topography such that projections from each cortical area terminate in longitudinal bands in the striatum (Selemon and Goldman-Rakic, 1985). Within the overall topography there is a more complex organization such that some reciprocally interconnected areas of frontal, temporal, and parietal cortex terminate in adjacent or interdigitating zones in the striatum (Selemonand Goldman-Rakic, 1985;Yeterian and Pandya, 1993;Yeterian and Van Hoesen, 1978). Although there is a general topographic relationship between cerebral cortex and striatum, there is also convergenceand divergenceof cortical inputs to striatum. The extent of convergenceand divergence

2. ANATOMY OF THE BASAL GANGLIA It usefulto considera overviewof the basal ganglia connections before discussing each connection in detail (Fig. 1). The basal ganglia may be viewed as two primary input structures, two primary output structures, and two intrinsic nuclei. The input structures of the basal ganglia are the striatum (caudate and putamen) and the subthalamic nucleus (STN). The striatum receivesexcitatory inputs from virtuallyall areas of cerebralcortex and STN receives excitatory inputs from motor areas of the frontal lobe. The output structures are the globus pallidus internal segment (GPi) and substantial nigra pars reticulate (SNpr). Both of these output structures receivea fast, divergent, excitatory input from STN and a slower, focused, convergent inhibitory input from the striatum. The outputs from GPi and SNpr are inhibitory and project to motor areas in the brainstem and thalamus. There are no direct outputs from the basal ganglia to spinal or brainstem motor neurons. The intrinsic nuclei are the globus pallidus pars externa (GPe) and substantial nigra pars compacta (SNpc). Like GPi, GPe receivesexcitatory input from STN and inhibitory input from striatum, but GPe sends inhibitory output to STN, GPi and SNpr. SNPC is the locus of dopaminecontaining neurons. It receives the bulk of its input from the

E@’h * Inhibitory

b

I

1

?

Brainstemand spinalcord -———..——

Fig. 1. Schematicrepresentationof basalgangliaconnections. GPe: globus pallidus,pars externa;GPi: globus pallidus,pars interna;IL: intralaminarthalamicnuclei; MEA:midbrainextrapyramidal area;SC:superiorcollicuIus;SNpe:substantial nigraparscompacta;SNpr:substantial nigra pars reticulate;STN: subthalamicnucleus;VA: ventral anterior thalamicnucleus;VLO:ventral lateral thalamicnucleus,parsoralis;DA:dopamine(withD, and DI receptorsubtypes);Enk:enkephalin;GABA:gammaamino-butyricacid; Glu: glutamate;SP: substanceP; ?: transmitterunknown.Open arrowheadsrepresentexcitatory projections,filled arrowheadsindicate inhibitory connections,half-filledarrowheads(Exe/Inh)represent projectionsthat can be excitatoryor inhibitory,depending on the receptorson the targetneurons.

384

J. W. Mink

Arm representation I

Cortex

I Area 4 I Area 3 I Area 1

Striatum

Globus Pallidus

The principal cell in the striatum is the medium spiny neuron. This cell type constitutes !)5°/0 of the striatal neuron population (Gerfen, 1988;Kemp and Powell, 1971). Medium spiny cells, have radial dendritic trees spanning up to 500 Km (Wilson and Groves, 1980).Thus, a singlemedium spiny cell may receiveinput from severalcortical areas. In addition to the cortical input, medium spiny striatal neruons receive a number of other inputs, including: (1) an excitatory and presumed glutamatergic input from the centromedian and parafascicular nuclei of the thalamus (Lapper and Bolam, 1992;Sadikot er al., 1992); (2) cholinergic input from the large aspiny striatal neurons (Izzo and Bolam, 1988); (3) y-amino-butyric acid (GABA), substance P, and enkephalininput from adjacent mediumspinystriatal neurons (Penny et af., 1986);(4) a large input from dopamine-containingneurons in the substantialnigra pars compacta (SNpc) (Carpenter, 1981). These inputs will be discussed below.

Fig. 2. Divergenceand convergencein the corticostriatal projection.AdaptedfromFlahertyand Graybiel(1994).

2.2. FocusedInhibitory Output from Striatum to Basal Ganglia Output Nuclei

in the sensorimotor corticostriatal projection has been studied in detail by Flaherty and Graybiel (Flaherty and Graybiel, 1991, 1993b, 1994). They injected small amounts of anterograde tracers into physiologicallyand somatotopicallyidentifiedsitesin somatosensory and motor cortex and confirmed previous observations that the projection from a single cortical area terminates divergently in several patches in the striatum (Flaherty and Graybiel, 1991; Selemon and Goldman-Rakic, 1985).By using two anterograde tracers, they were able to extend these observations in two regards. First, there is a preservation of somatotopy such that striatal zones receivinginput from the facearea of sensoryor motor cortex are separate from those receivinginput from the arm area. Second, there is convergenceof inputs from more than one cortical area such that the projections from the face areas of motor and sensory cortex overlap (Flaherty and Graybiel, 1991).Thus, for the somatosensoryand motor inputs to striatum, there is both a “one-to-many” and a “many-to-one” pattern of connectivity (Fig. 2). Using a similar technique, GraybieI’sgroup has shown that a similar pattern of convergenceand divergenceexists for the projection from frontal eye fieldsand supplementary eye fields to striatum (Parthasarthy et al., 1992). Projections from somatic premotor areas (supplementary and arcuate motor areas) adjacent to the oculomotor areas also convergedwith one another in striatal zones that were separate from, but adjacent to, the targets of the oculomotor projections. Despite the extensiveoverlap, in no instance was the overlap of cortical projections to striatum complete, leaving open the possibility that a small region of striatum could receive a private input from a single cortical area. The multiply convergent and divergent pattern of the corticostriatal projection provides an anatomical substrate in the striatum for the integration of information from several different areas of cerebral cortex (Graybiel et al., 1994).

Medium spiny cells are the striatal output neurons and project to globus pallidus and substantialnigra (Gerfen, 1988; Kemp and Powell, 1971). Thus, approximately $J5°/0 of Striatalneurons project out of the striatum. Medium spiny neurons contain the inhibitory neurotransmitter GABA (Ribak et al., 1979) and colocaltied peptide neurotransmitters (Penny et al., 1986).Based on their neurotransmitter content and post-synaptic target, medium spiny neurons can be divided into three populations. One population contains GABA, dynorphin, and substance P and sends axons to GPi or SNpr (Albin et aZ., 1989;Gerfen et al., 1990;Gerfen and Young, 1988).The second population contains GABA and enkephalin and projects to GPe. These two populations of medium spiny neurons are morphologically indistinguishable and are not topographically segregated within the striatum (Feger and Crossman, 1984; Gerfen and Young, 1988; Parent et al., 1989). The commingling of these output neurons suggeststhat they receivesimilar inputs and conveysimilarinformation to their respectivetargets. The third population of medium spiny cells contains GABA, substance P and dynorphin and projects to substantialnigra pars compacta (Gerfen and Young, 1988).This latter population is segregated anatomicallyfrom the former two and willbe discussedbelow (section 2.3). The termination pattern of striatal projections to GPi, SNpr, and GPe suggestsa focused influenceof striatum on these structures. It has been shown with anterograde tracers that striatal axons entering GPi or SNpr contact several neurons in passing before ensheathinga singleneuron with a dense termination (Fig. 3) (Parent and Hazrati, 1993).While a single striatal axon may contact severaltarget neurons, this termination pattern suggests that the physiological effect is more focused. Two lines of evidencesuggestthat striatal input to GPi is convergent. First, the dendrites of an individual GPi cell span up to 1 mm in diameter, organized in a disc-like tree that is oriented

TheBasalGanglia perpendicular to incoming striatal axons (Percheron et al., 1984).This arrangement would maximize the potential for convergenceof striatal inputs. Second, when a single injection of retrograde tracer is made into GPi, several zones are labeled in striatum (Flaherty and Graybiel, 1993a;Gimenez-Amayaand Graybiel, 1990, 1991). The striatal zones appear similar to those that are labeled with injection of anterograde tracers into cerebral cortex (section2.1). When Flaherty and Graybiel (1994) injected retrograde tracer into GPi and anterograde tracer into sensory or motor cortex they were able to demonstrate multiple striatal zones that were labeled from both injections. This suggeststhat information is sent from cortex to striatum in a multiply convergentand divergentpattern with reconvergence in GPi after processing in the striatum (Fig. 2). Other evidence suggests that striatal neurons project to GPi and SNpr in segregated parallel pathways. In the monkey, retrograde double tracer studies have shown that the bulk of the caudate nucleusprojects to SNpr and the putamen projects to GPi (Hedreenand DeLong, 1991;Parent et al., 1984). When anterograde tracers were injected into two nearby but non-adjacent sites in the putamen, there was little overlap of their termination zones in GPi (Hazrati and Parent, 1992b). The topography and apparent lack of convergence of projections from striatum has been emphasized in a model of parallel

385

distributed processingin the basal ganglia (Alexander et al., 1986).One way to reconcilethe viewsof those who emphasizeparallelism(Alexanderand Crutcher, 1990a, 1991;Alexander et al., 1986)and those who emphasizeconvergence(Percheron and Filion, 1991; Percheron et af., 1984)is to conclude that there are parallel projections to GPi and SNpr from widely separated areas of striatum, but convergent projections from nearby and functionally related areas of striatum. However,this type of organization remains to be proven. The organization of the striatal projection to GPe is generally similar to the projections to GPi and SNpr (Flaherty and Graybiel, 1993a; Hazrati and Parent, 1992a, 1992b;Yelnik et al., 1984). The output of the striatum is inhibitory to all targets. Electrophysiologicalstudies in anesthetized rats or in rat brain sliceshave shownthat stimulation of the striatum results in relatively long latency inhibition in SNpr, entopeduncular nucleus (the rodent homologue of GPi), and GPe (Nakanishi et al., 1987, 1991; Robledo and Feger, 1990).This inhibition appears to mediated by GABA (Kita and Kitai, 1988;Precht and Yoshida, 1971). 2.3. HeterogeneousIntrinsic Organization of the Striatum

There are no apparent regional differencesin the striatum based on cell morphology. However, an intricate internal organization has been revealedwith special stains. When the striatum is stained for acetylcholinesterase (AChE), there is a patchy distribution of lightly staining regions within more heavily stained regions (Graybiel and Ragsdale, 1978). The AChE-poor patches have been called striosomes and the AChE-rich areas have been called the extrastriosomal matrix. The matrix forms the bulk of the striatal volume and receivesinput from most areas of cerebral cortex. Within the matrix are clustersof neurons that have been termed matrisomes (Graybielet al., 1994).The matrisomescorrespond to the zones of termination of cortical afferents and to zones that are labeled with retrograde tracers in the pallidum. The output from cells in the matrix is to both segments of the GP and to SNpr as discussed (section 2.2). Striosomes receive input from GPi above limbicand prefrontal cortex and send output to SNpc (Gerfen, 1992). Immunohistochemical techniques have demonstrated that many peptides such as substance P, dynorphin, and enkephalin have a patchy distribution that may be partly or wholly in registerwith the striosomesor matrix (Graybiel et al., 1981). The striosome-matrixcompartmental organization confers upon on the striatum a potential functional \ segregation. The dendrites of most medium spiny \ neurons are restricted to a single compartment, indicating relative privacy of inputs to each compartment (Walker et al., 1993). Both compartments receive input from the dopamine neurons of Fig. 3. Divergentexcitatoryprojectionfrom subthalamic the substantialnigra pars compacta (SNPC),but only nucleusand focusedinhibitoryprojectionfromstriatumto the striosomesproject back to the SNpc. Despite the globuspallidusinternalsegment.Abbreviations— GPI: segregation,there is evidencethat the striosomescan globuspallidusinternalsegment;STN:subthalamicnucleus affectthe matrix via large aspiny interneurons. These cell make up 1–2°/0of the striatal population and (adaptedfromParentand Hazrati(1993)).

()

STN

386

J. W. Mink

contain the neurotransmitter acetylcholine (ACh). The dendrites of these cholinergic neurons extend into both striosomes and matrix and their axons ramify mostly in the matrix (Kawaguchi, 1992).The functional consequences of the striosome-matrix organization are still largely unknown.

hyperpolarized state ( – 80 mV) (Jiang and North, 1991;Kawaguchi et al., 1989).This inward rectifying potassium current acts to shunt weak inputs. It has been suggested that relatively coherent excitatory input over much of the dendritic tree is required to overcome this inward rectifying current (Wilson, 1995).As the cellis depolarized,the inward rectifying current is inactivated and if excitatory inputs are 2.4. Mechanismsthat Focus Striatal Activity stifficientlypowerful and coherent, the membrane The compartmental organization and pattern of potential rapidly depolarizesto a point at which it is termination of cortical inputs provide an anatomical dominated by outwardly rectifying potassium curbasis for focusing information in the striatum. rents (Bargas et al., 1989; Surmeier et al., 1991). Several other mechanisms may also contribute to These outwardly rectifyingcurrents tend to maintain focusing, including the electrophysiologyand mor- the resting potential in a relatively depolarized, but phology of medium spiny neurons, the dopamine still subthreshold state ( – 50 mV). This depolarized input from SNpc, interactions betweenmediumspiny state is near threshold, so that small fluctuations of neurons; and striatal interneurons. Together, these the membrane potential in response to inputs can several mechanisms appear to result in an output trigger action potentials. The depolarized state is signal from the striatum that is spatially and maintained until the level of excitatory input temporally focusedand that is modifiablein relation decreases and the inward rectifying current is to behavioral contexts. reactivated, returning the celI to a hyperpolarized state (Fig. 4). Based on the membrane properties, the generation 2.4.1. Spatial and Temporal Focusing by Medium of action potentials in a medium spiny neuron would Spiny Striatal Neurons require coincidentexcitatory input to many dendritic As noted above, approximately !)5~0 of striatal spines. Becauseof the small number of contacts onto neurons are medium spinyprojection neurons (Kemp a medium spiny neuron from a singlecortical axon, and Powell, 1971).A typical medium spiny neuron this would’require the activation of a large number has approximately 2&60 dendrites that branch span of corticostriatal neurons. Thus, activation of a 200 to 500 ~m (Wilson and Groves, 1980).Wilson medium spiny neuron would require spatial and (1995) has estimated that each spiny dendrite temporal convergenceof cortical inputs resulting in contains about 500spines,about half of whichreceive the activation of discrete foci in striatum in response a single synapse from a cortical afferent fiber. Based to specificpatterns of cortical input. on the number of dendrites and dendritic spines, it has been estimated that the typical medium spiny 2.4.2. Dopamine Input from Substantial Nigra Pars neuron receives 5000 to 15,000 cortical synapses Compacta Modulates Cortical Inputs (Wilson, 1995). Each corticostriatal fiber contacts Few neural systemshave attracted as much interest several striatal neurons that are spread out over a relatively large region of the striatum, but a typical as the nigrostriatal dopamine system. It is the corticostriatal fiber makes very few contacts on an dopaminergiccells in SNpc that are most vulnerable individualmedium spinyneurcm(Cowanand Wilson, in Parkinson’s disease (Adams and Victor, 1993). 1994).This anatomical finding is consistent with the Furthermore, a largebody of work has implicatedthe observation of small EPSPSin mediumspinyneurons dopaminergic system in mechanisms of reward and (Wilson et al., 1982).Wilson (1995) has estimated reinforcement (see Wickens and Kotter (1995) for that one striatal cell receivesinput from thousands of review). Despite the interest and much work, the cortical neurons. Based on these tindings, it is exact role of the dopaminergic input to striatum is unlikely that the activity of a given medium spiny still debated. The dopamine input to the striatum terminates neuron is controlled by input from any one cortical neuron. Rather, activation of a mediumspinyneuron primarily on the shafts of the dendritic spines of appears to require concurrent excitatory input from medium spiny neurons (Bouyer et al., 1984)(Fig. 5). Sincethe cortical afferentsterminate on the heads of a large number of cortical afferents. The membrane properties of medium spiny spines, the dopamine input is in a position to the neurons also contribute to focusing in the striatum. modulate the influenceof cortical inputs. It has been It has long been known that medium spiny neurons suggestedthat the dendritic spine is a chemicallyand have a very low rate of spontaneous discharge, but electrically segregated compartment that effectively until recently the underlying mechanism was not localizes the effect of dopamine on each cortical known. It had previouslybeen suggestedthat the low input. Although the interaction may be locally discharge rate was due to tonic inhibitory inputs to isolated, a singledopaminergicaxon may synapse at the medium spiny neurons. Whilethere is anatomical severallocations on a singledendrite and on multiple evidencefor input from inhibitorycollateralsof other dendrites of multiple striatal neurons (Groves et al., medium spiny neurons, physiological evidence has 1994).Thus, activity in a single SNpc neuron may shown these inhibitory inputs contribute little to the influencemany cortical inputs to many mediumspiny resting membrane state (Wilson, 1995).The resting striatal neurons and any specificity of dopamine potential of the medium spiny neurons is dominated modulation of corticostriatal inputs is likely to by an inwardly rectifying potassium current that depend on the temporal relationship betweencortical tends to keep the membrane potential in a and nigral activity.

TheBasalGanglia

I

387

g/gmax 1 #

o Vm(mV) ‘ o-20-

20mV

IAs

I

-50mV +

Jb

+-S0

mV

1

h#? w

0.5S@c

-40-60-so -1oo -120

i,

1 “) ~,,, inward rectifier ------

Fig. 4. Two stablemembranestates in mediumspinystriatal neurons.Mediumspinyneuronsare maintainedin a hyperpolarized state( —80mV)byan inwardrectifierpotassiumcurrentandin a more depolarizedstate ( – SOmV) by an outward potassiumcurrent (1.s). At membranepotentiak < – – 70mV,theinwardrectifierconductance(g)increasesandat potentials> –50 mV,theoutward potassiumconductanceincreases(fromWilson(1995);reproducedwithpermissionof MIT Press).

modulatory mechanisms described below, temporal coincidence of these reward-predicting signals with cortical input to medium spiny neurons would increase the synaptic strength of the cortical inputs. The strengthening of the temporally coincident cortical input would increase the probability of the mediumspinycellfiringthe next time it receivesthose sameinputs. Sincea givenmediumspiny neuron may receivemultiple patterns of input, the strengthening of some patterns of input and weakening of others would sculpt the activity pattern of a medium spiny interneuron cell such that it would firemore in some contexts and less in others. This idea of increasing or decreasing synaptic strength by concurrent activity of two differentinputs is an old one that has been applied to neurons models of learning in other systems (Hebb, 1949; Morris et al., 1988;Thach et al., 1992). The action of dopamine on striatal neurons [V \ &.bsta.Ce~ depends on the type of dopamine receptor involved. Five types of G protein-coupleddopamine receptors have been describedand have been grouped into two families, D1 and D2, based on their response to agonists (Sibley and Monsma, 1992).It is generally agreed that D1 receptors stimulate and D2 receptors inhibit adenyl cyclaseactivity. There is also evidence ~ Glutamate carboxylase that dopamine transmission influences the conduc,~-,=’”,--””” f (GABA) tance of sodium, potassium, and calcium ions (Surrneier et al., 1993), but it appears that these effectsare primarily local and have little influenceon -output the membrane potential of the entire cell. Both D1 Fig. 5. Anatomicalrelationshipof afferent inputs to and D2 receptors are located on medium spiny cells. mediumspiny striatal neurons(from Smithand Bolam Some evidence suggests that D1 receptors are (1990);reprintedwithpermissionof EleevierScienceLtd.). preferentiallylocated on cells that project to GPi or

As willbe discussedbelow(section3.8),the activity of nigrostriatal dopamine neurons correlates with reward or with anticipation of reward. Acting via

388

J. W. Mink

SNpr and that D2 receptors are located on cells that project to GPe (Gerfen e~al., 1990).However, there is debate as to whether cellswith D1 or D2 receptors are completely separate populations, or whether the receptors are colocalized on some cells (Surmeier et al., 1993). In addition to the postsynaptic localization, D2 receptors are also present presynaptically on corticostriatal terminals and on other dopaminergicterminals (autoreceptors)(Kalsner and Westfall, 1990). Acting via both pre- and post-synaptic receptors, dopamine can modulate corticostriatal transmission by mechanisms of long-term depression (LTD) and long-term potentiation (LTP) (Groves et al., 1995). Through these mechanisms,dopamine strengthensor weakens the efficacyof corticostriatal synapses.The effect of LTP and LTD depends on temporal coincidence of cortical and nigral inputs to the striatum. This interaction contributes to the temporal focusing of striatal output. If learning does modifiy activity in the striatum, what is learned?To answer that question, we need to understand the functions of the basal ganglia since what is learned is likely to be specific to these functions. One possibility is that striatal activity is modified by learning in order to appropriately select desired motor mechanisms and inhibit others in the context of ongoing posture and movement. This hypothesis will be discussed in section 5.

2.4.4. Cholinergic Input Regulates the Membrane State of Medium Spiny Neurons

Large aspiny striatal neurons are interneuronsthat use acetylcholine (ACh) as a neurotransmitter (Phelps et al., 1985). They have extensive axon collaterals in the striatum that terminate on medium spiny neurons. Based on studies correlating extracelhdar discharge characteristics with anatomical localization, it appears that these neurons have a distinctive action potential profile and a tonic discharge rate (Aosaki et al., 1995). Intracellular recording has shown that ACh stabilizes the membranepotential of medium spinyneurons (Akins et al., 1990).Recall that medium spiny neurons have bistable membrane potentials that alternate between a hyperpolarized state and a depolarized state. Through their stabilizingeffect,activityof cholinergic aspiny neurons would cause hyperpolarized cells to remain insensitiveto cortical inputs and depolarized cells to remain responsiveto cortical inputs. The net effect over several repetitions would be to enhance medium spiny neuronal discharge patterns that are associated with cholinergic activity. The cholinergic input likelyinteracts with dopamine input (Di Chiara et al., 1994)to strengthen dischargepatterns that are associatedwith desired behavioral outcomes (Aosaki et al., 1994). 2.5. Excitatory Input from Motor Areas of Cerebral Cortex to Subtbalamic Nucleus

2.4.3. Medium Spiny Neurons Have Extensive Local Axon Collaterals

STN receives a short-latency, excitatory, glutamatergic input from motor areas of ipsilateral Studies of intracellularly stained medium spiny cerebral cortex includingprimary motor cortex (area neurons have revealed extensive intrastriatal axonal 4), premotor and supplementary motor cortex (area arbors in addition to their axonal projection 6), and frontal eyefields(area 8) (Fujimoto and Kita, outside of the striatum (Kawaguchi et al., 1990; 1993; Hartmann-von Monakow et al., 1978; Wilson and Groves, 1980). Two patterns of Rouzaire-Duboisand Scarnati, 1987).The projection arborization have been identified. The more com- from motor cortex (area 4) is somatotopically mon type ends in a field approximately 400 ~m in organizedand is denser than that from more anterior diameter, roughly overlapping with the dendritic areas (Hartmann-von Monakow et al., 1978).The field of the cell of origin. The other type covers a cortico-subthalamic projection appears to be topofield over 1 mm in diameter and extends beyond its graphicallyorganized(Afsharpour, 1985;Hartmannown dendritic field. (Kawaguchi et al., 1990).Some von Monakow et al., 1978). However, the dendrites authors have drawn the inference that this anatom- of STN neurons extend up to 1200pm and so there ical arrangement provides the framework for may be considerableconvergenceof inputs from the recurrent or collateral inhibition (Park et al., 1980). different cortical areas onto an individual neuron Electron microscopic evidence indicates that (Yelnikand Percheron, 1979).The STN also receives medium spiny neurons do indeed form synapses an inhibitory GABA input from GPe which will be with neighboring medium spiny neurons (Wilson discussed below (section 2.9.1). and Groves, 1980)and immunocytochemicalstudies have shown that these synapses are likely to use 2.6. Divergent Excitatory Input from Subtbalamic GABA (Ribak et al., 1979). When applied to Nucleus to Basal Ganglia Output Nuclei medium spiny neurons, GABA causes a short The output from the STN is excitatory and acting inhibition that appears to be mediated by chloride (Lighthall and Kitai, 1983).Yet, there has glutamatergic and projects to GPi, GPe, and SNpr. never been any direct physiologic evidence of The connections with GPe will be discussed below collateral inhibition among medium spiny neurons (section2.9.1).The projectionsfrom STN to GPi and (Jaeger et al., 1994),and it has been suggestedthat SNpr dispIay a rough topography (Smith et al., intrastriatal inhibition is mediated via GABAergic 1990).In the monkey, the majority of the projections interneurons (Wilson et al., 1989).At this time the from STN to GPi and from STN to SNpr arise from role of the extensive local collaterals is not known. separate populations of neurons (Parent and Smith, Collateral inhibition among medium spiny neurons 1987b). In GPi, STN fibers form bands that are is a theoretical, but not proven, mechanism of parallel to the internal medullary Iamina and in focusing in the striatum. register with the disc-like dendritic fields of the

TheBasalGanglia pallidal neurons (Smithet al., 1990).In both GPi and SNpr, STN axon collaterals branch to ensheathe the cell bodies and proximal dendrites of their target neurons (Hazrati and Parent, 1992a).The STN to GPi projection appears to be highly divergent such that each axon from the subthalamic nucleus ensheathes many GPi neurons (Parent and Hazrati, 1993).This is in contrast to the more focused input from striatum (Fig. 3). Historically the output of STN was thought to be inhibitory (Carpenter, 1981), but there is now strong evidencethat it is excitatory and glutamatergic(Brotchie et al., 1991a;Rinvik and Ottersen, 1993).Neurons in the rat entopeduncular nucleus respond to STN with a short latency excitation (Nakanishi er al., 1987,1991;Robledo and Feger, 1990)that is blocked with application of the glutamate antagonist kynurenate (Robledo and Feger, 1990). 2.7. Critical Differences Between the Subthalamic and Striatal Outputs

Although less is known about the organization of the STN than the striatum, it is clear that several important differencesexist between the outputs of these two structures. 1. STN receivesexcitatory input from OCU1Oand somato-motor areas of the frontal lobes;the striatum receivesexcitatory input from virtually all of cerebral cortex. 2. STN appears to have little intrinsic processing of transmitted cortical signals; the striatum has multiple mechanisms for intrinsic processing and transformation of information. 3. STN sends a fast, divergent,excitatory signalto GPi and SNpr; striatum sends a slower, more focused, inhibitory signal to GPi and SNpr (Fig. 3). The net effectof this arrangement is broad excitation of GPi and SNpr in response to motor commands from cerebral cortex and focused inhibition of specific GPi or SNpr neurons in a behaviorally specific,context dependent manner. 2.8. Globus Pallidus Para Interna and Substantial Nigra Pars Reticulate are Output Nuclei GPi and SNpr are composed of large neurons that receivesimilar patterns of input. On histologicaland connectional grounds, it has been suggested that these two nuclei should be considered to be one structure that is divided during development by the internal capsule (Carpenter, 1981).Supporting data come from physiologicalstudies of somatotopy that have shown that the body below the neck is represented in GPi and that the face and eyes are reprewnted in SNpr (DeLong et al., 1983). This section will focus on the outputs of these nuclei. 2.8.1. Inhibitory Output from Globus Pallidus Pars Interna to Thalamus and Brainstem

In primates, the globus pallidus is divided into internal and external segments by the internal medullary lamina. In rodents and carnivores, the internal segment lies within the internal capsule and

389

is called the entopeduncular nucleus. The external segment of the globus pallidus (GPe) is not a principal sourceof output from the basal ganglia and is considered below (section 2.9.1). The GPi is primarily composed of large neurons that project outside of the basal ganglia (DiFiglia and Rafols, 1988; Francois et al., 1984). These neurons are inhibitory and use GABA as a neurotransmitter (Penney and Young, 1981). Based on retrograde double labeling, it appears that about 70~o of GPi neurons send collaterals to both thalamus and brainstem (Parent and De Bellefeuille,1982).Other GPi neurons (20?4.)project to the centromedianparafascicular complex of the thalamus or to the lateral habenula (Parent and De Bellefeuille,1982). The role of these projections is not known. In the thalamus, most axons from GPi terminate in the oral part of the ventrolateral nucleus (VLO)and in the principal part of the ventral anterior nucleus (VApc) (DeVito and Anderson, 1982). These thalamic nuclei project to motor, premotor, supplementary motor, and possibly prefrontal cortex (Hoover and Strick, 1993; Inase and Tanji, 1995; Middleton and Strick, 1994;Tokuno et al., 1992). These cortical areas also receive input from other thalamic nuclei that receive cerebella input. However, there is little or no overlap of the basal ganglia and cerebella inputs to thalamus (Rouiller et al., 1994;Yamamoto et al., 1983).Therefore, the basal ganglia and cerebellum each have private lines via thalamus to the motor areas of the frontal lobes. There is evidencethat an individualGPi neuron sends output via thalamus to just one area of cortex (Hoover and Strick, 1993).GPi neurons that project to motor cortex are adjacent to, but separate from, those that project to premotor cortex. This arrangement is evidencefor functionally segregatedparallel outputs of the basal ganglia (Alexander et al., 1986; Hoover and Strick, 1993). Collaterals of most GPi axons projecting to thalamus project to an area at the junction of the midbrain and pens adjacent to the pedunculopontine nucleus (PPN). The terminology for this area is disputed. Some authors suggestthat the target of the GPi efferents is the PPN proper (see Lavoie and Parent (1994)),while others have suggestedthat it is separate from the cholinergic neurons of the PPN (Rye et al., 1988).The latter group of authors has coined the term “midbrain extrapyramidal area” (MEA) to identify the target of GPi efferentsin this region.The midbrain extrapyramidalarea projects in turn to the reticulospinal motor system. 2.8.2. Inhibitory Output from Substantial Nigra Pars Reticulate to Thalamus, Superior Colliculus and Brainstem

Like the globus pallidus, the substantialnigra is dividedinto two segments.One is a denselycelh.darly regioncalledthe pars compacta and the other is more sparsely cellular and is called the pars reticulate (Carpenter, 1981). The pars compacta contains dopamine cells and is not a principal output nucleus of the basal ganglia. It will be considered below (section 2.9.2). Like GPi, the substantial nigra pars reticulate

390

J. W.Mink

(SNpr) contains large neurons that project outside of the basal ganglia. The output is GABAergic and inhibitory to the medial part of ventrolateral thalamus (VLm) and to the magnocellular part of ventral anterior thalamus (VAmc)(Carpenter et al., 1976;Oertel and Mugnaini, 1984;Ueki, 1983).These thalamic areas in turn project to premotor and prefrontal cortex (Kievit and Kuypers, 1977).Like the GPi output, SNpr sends collaterals to the midbrain extrapyramidal area and also sends a projection to the centromedian-parafascicularcomplex of the thalamus (Francois et al., 1988; von Krosigk et al., 1992).The primary differencebetween the outputs of GPi and SNpr is that the lateral portion of SNpr is connected with cortical and brainstem areas that control eye movements. This lateral part of SNpr sends an inhibitory projection to the superior colliculusand to the paralaminar part of the dorsal medial thalamus (DMpl) (Carpenter er al., 1976; Hikosaka and Wurtz, 1983d; Rinvik et al., 1976).DMpl projects in turn to the frontal eye fields.

of caudate and putamen in a topographic manner (Hedreen and DeLong, 1991).The pattern and effect of the dopamine input to striatum was discussed above (section 2.4.2). Recent evidence has shown that STN (Campbell et al., 1985)and GPi (Parent and Smith, 1987a)also receivea dopamine input. While these inputs are still not well characterized, they may be of significant importance to the function of the basal ganglia. 2.10. Basal Ganglia Anatomy: Summary and Synthesis

1. The striatum receivesinput from nearly all of cerebral cortex such that several functionally related cortical areas project to overlapping striatal zones and that an individualcortical area projectsto several striatal zones. Cortical areas that are not functionally related project to separate zones of the striatum, although there may be some striatal neurons that receiveinput from more than one adjacent zone. 2. There are multiple mechanisms intrinsic to the striatum that integrate inputs and focus the output. 2.9. Globus Pallidus Pars Extema and Substantial 3. The striatum sends a focused inhibitory Nigra Pars Compacta projectionto the basal gangliaoutput nuclei,GPi and Globus pallidus pars externa (GPe) and substantial SNpr. It also sends an inhibitory projection to GPe nigra pars compacta (SNpc) may be viewed as which, in turn, inhibits STN and GPi. This latter intrinsic nuclei of the basal ganglia. That is, they “indirect” pathway from striatum through GPe (and receivethe bulk of their input from and send the bulk STN) could act in opposition to the direct pathway of their output to other basal ganglia nuclei. and result in further focusingof the information flow from striatum to GPi. 4. The subthalamic nucleus receives input from 2.9.1. Globus PaUidus Pars Externa Receives Input motor, premotor, and supplementary motor cortex from Striatum and STN and Sends Output to STN, and from the frontal eye fields. GPi and SNpr 5. The subthalamic nucleus sends a fast divergent The inputs to GPe are similar to inputs to GPi. It excitatory projection to GPi and SNpr. receives an inhibitory projection from the striatum 6. The output from the basal ganglia (GPi and and an excitatory one from STN. The pattern of SNpr) is inhibitory and projects to motor and termination of the striatal and STN afferentsin GPe premotor areas in the brainstem and thalamus. There is similar to GPi: the striatal input is focused and are no direct connections between the basal ganglia relatively discrete while the STN input is divergent and spinal sensory or motor neurons. (Parent and Hazrati, 1993).The output is GABAergic and inhibitory and the majority of the output projects to STN (Rouzaire-Dubois et al., 1980).In addition, 3. ACTIVITY OF THE BASAL GANGLIA there is a recently described GABAergic inhibitory DURING MOVEMENT output from GPe directlyto GPi and to SNpr (Bolam Although the anatomic organization of the basal and Smith, 1992).SinceGPe and GPi get input from anatomically intermixed striatal neurons they are ganglia providesthe infrastructure for their function, likely to receive similar information. The fact that any inferenceof function from anatomy can only be GPe inhibits GPi directly or via STN suggests that speculative. One way to obtain direct information GPe may act to oppose, limit, or focus the effect of about the function of a structure is to measure its the striatal projection to GPi (Alexander and activity during behavior. Severalmethods have been used to measure the activity of differentcomponents Crutcher, 1990a). of the basal ganglia including single unit recording, 2-deoxyglucose (2-DG) autoradiography, and 2.9.2. Substantial Nigra Pars Compacta is positron emission tomography (PET). The scope of Reciprocally Connected with the Striatum this reviewwillbe limitedprimarilyto resultsof single The SNpc is made up of large dopamine-contain- unit recordingduring behaviorin awake animals. For ing cells. SNpc receives input from the striatum, reviewsof 2-DG and PET studiesof the basal ganglia specifically from the striosomes (Graybiel, 1990). during behavior, see Brooks, 1995) and Crossman This input is GABAergicand inhibitory.Other inputs (1987). Singleunit recordingoffersseveraladvantagesover to SNpc have been difficult to assess because the dendrites of SNpc and”SNpr neurons overlap and it the other techniques. First, it is the only direct way is not always clear whether individual axons ending to measure neuronal activity during behavior. in the substantialnigra terminate on SNPCcells,SNpr Second, it allows the neuronal activity to be cells, or both. SNPCdopamine neurons project to all correlated with behavior with a temporal resolution

TheBasalGanglia

391

of milliseconds.Thus, the activity can be related to (Crutcher and DeLong, 1984b;DeLong, 1973).The severalindividualcomponents of a task. By contrast, types of neuronal activityin the putamen during limb the other techniques have a temporal resolution on movement can be divided into several groups. One the order of seconds to minutes which requires pattern of activity is time-locked to movement and multiple repetitions of a highly stereotyped behavior occurs during the movement. A second pattern for a singlemeasurement.Third, singleunit recording followsan instructional cue and precedesmovement. allowsa spatial resolution of fractions of a millimeter This type of activity is thought to be related to the whereas the other techniques provide resolution of intent or “set” of the animal and is often referred to severalmillimeters.The chieflimitation of singleunit as set-related activity. A third type of activity is recordingis the need to record a large enough sample related to movementoccurringin the context of prior of units to get a reliable representation of the entire movementsor specifictask conditions. Finally, some population in a given area. Even with large samples, striatal neurons fire in relation to specific sensory there is still a bias toward recording large neurons stimuli, usually when the stimulusis presented in the more readily than small ones. context of a movement. Despite the promise of sampling neuronal activity during behavior, this may not be enoughto determine 3.1.1. Movement-Related Striatal Neurons are the function of a structure. Great care must be taken Somatotopically Organized to assure that an apparent correlation of brain activity with a given variable is specific for that Striatal neurons with movement-related activity variable and not correlated with other more covert have a distinct somatotopic organization with the variables. One must be aware that a lack of a phasic face representedventromediallyand the leg dorsolatchange in neuronal activity during a task does not erally in the middle and caudal regions of the necessarilymean that structure is uninvolvedin that putamen (Crutcher and DeLong, 1984a).It is these task. A structure may play a critical role in the regions that receive input from somatosensory and performanceof a task by maintaining a tonic levelof motor cortex (Flaherty and Graybiel, 1991, 1993b; activity. Finally, it must be remembered that Kunzle, 1975). When recorded with extracellular correlation does not imply causation and other microelectrodes, neurons tend to occur in clusters techniques are needed to confirm inferred causal with similar dischargecharacteristics(Alexanderand relationships of brain activity to specificbehaviors. DeLong, 1985).Microstimulationin the area of these Nevertheless, with well designed experiments and clusters evokes movement of the body part with careful data analysis, useful functional information which nearby neuronal dischargecorrelates (Alexander and DeLong, 198’5).These functionallyidentified can be obtained. DeLong (1971)was the first to demonstrate that clusters may correspond to the anatomically defined neuronal activity in the basal ganglia is correlated matrisomes described above (section 2.4). If arm movementsare made with constant torque with movement. Since that pioneering work, there has been an increasing effort to identify just what loads either opposing or assistingthe movement, the aspects of movement basal ganglia neurons seem to direction of movementcan be dissociatedfrom which code. Neurons in the different basal ganglia nuclei muscles are used to make the movement. For have characteristic baseline discharge patterns that example,a flexionmovementcan be made against an change with movement. In trained animals, the opposing load by turning on the flexors or with an activity of single neurons can be studied in relation assisting load by turning off the extensors. When to different aspects of movement including prep- putamen neurons were recorded in this type of task, aration and executionof movement,parameters such approximately f@O/o of movement-related putamen as direction, amplitude, or velocity, and modes of neuronsfiredin relation to the direction of movement guidance. In addition to measuringthe correlation of and 25°/0 fired in relation to the pattern of muscle neuronal activitywith specificvariabIes,the timing of activity(Crutcher and Alexander, 1990;Crutcher and neural activity in relation to an event can be DeLong, 1984b).Forty-one percent of movement-redetermined. The timing in the basal ganglia can then lated putamen neurons also had somatosensory be compared to the timing in other brain structures responses, most of which were related to joint to determine a relative sequenceof neuronal activity rotation. The activity related to passivejoint rotation was often correlated to the direction of the in a particular movement. movement. Apart from movement direction and muscle 3.1. Neuronal Activity in the Putamen Correlates pattern, correlation of putamen discharge with with Limb Movement physical parameters of movement (e.g. velocity, Most studies of neuronal activity in the basal amplitude, force, position, acceleration)has not been ganglia during movement have focused on the demonstrated. One early report suggestedthat more striatum and on the putamen in particular. The putamen neurons fire during slow “ramp” movemajority of striatal neurons have low baseline ments than during fast “ballistic” movements dischargerates of 0.1 to 1spike/second(Crutcher and (DeLong, 1973).However,it was found subsequently DeLong, 1984a; DeLong, 1973; DeLong and that more proximal muscles were active in the Georgopoulos, 1981).These neurons project outside “ramp” movement than in the “ballistic”. Thus, of the striatum and are most likelymediumspinycells putamen neurons that appeared to relate to slow (Kimura et al., 1990). When the activity of these movements may have been related to proximal striatal neurons is related to movement,the change is muscle activity instead (DeLong and Georgopoulos, usually an increase from the low baseline rate 1981).

392

J. W.Mink

3.1.2. Movement-Related Striatal Neurons Fire after Movement Has Been Initiated When neurons fireis as important to understanding the function of a structure as is what variables they may code. If a brain structure is responsiblefor the initiation of movement, the neurons in that structure must act before the onset of activity in musclesused to make that movement. In the putamen, most movement-related activity is late. In one study, the average time of putamen firing was 20 msec prior to elbowmovementonset and 50 msecafter the onset of EMG in the agonist muscle (Crutcher and DeLong, 1984b). In another study the average time of movement-related activity in the putamen was 33 msecafter the onset of elbowmovement(Crutcher and Alexander, 1990).In this latter study, the timing putamen activity was compared to motor cortex and supplementary motor area (SMA) in the same animals and was found to occur on average 56 msec after motor cortex and 80 msec after SMA. In anterior areas of putamen, some neurons discharge relatively earlier in relation to arm movement.In a reachingtask, approximately30Y0of neurons recorded in anterior putamen and caudate fired >140 msec prior to movement (Romo et al., 1992).In that study, 37y0 of anterior striata] neurons fired less than 140msec prior to movement(after the earliest EMG), and 33°10fired after the onset of movement. In another study, some putamen neurons fired best in relation to the stop rather than the start of movement (Montgomery and Buchholz, 1991). Thus, it appears that the timing of movement-related putamen activityis distributed over a widerange with a tendency for anteriorly located cells to fire earlier than posterior cells. Even with the broad timing distribution, the majority of both anterior and posterior movement-relatedcellsfire after movement has been initiated.

3.1.3. Some Striatal Neurons Have “Set” Signals

direction to move the forearm. In that task, 80% of putamen neurons had movement-relatedactivity and 23y0 had set-related activity. Of the neurons with set-related activity, most (79%) were directionally selective. In a subsequent study, Alexander and Crutcher, 1990b)compared set-related activity in the putamen to set-related activity in motor cortex and SMA. They found that qs~. of task-related putamen neurons had “set’’-related activity compared with 37% in motor cortex and 55% in SMA. In a task involving a whole arm reach, 30°/0 of putamen neurons had “set’’-related activity and 500/o of these weredirectionallyselective(Jaegeret al., 1993).Some neurons with set-relatedactivity also had movementrelated activity, but on average putamen neurons with “set’’-related activity were located anterior to those with movement-related activity. The more anterior areas of striatum receive inputs from prefrontal, premotor, and supplementary motor cortex (Flaherty and Graybiel, 1993b;Goldman and Nauta, 1977). The directional “set’’-related activity described above occurred in a tasks in which direction was the instructed variable. In another seriesof experiments, Schultz and his colleaguestrained monkeys to reach to a target in a “go-no go” paradigm (Schultz and Romo, 1992).In that task, the monkeywas instructed either to move or not to move in response to a subsequent trigger signal. In the “go” trials, the monkeyreached for a morsel of food; in the “no go” trials the monkeydid not move. Approximately20’Yo of neurons in the anterior putamen and caudate discharged in relation to the go-no go instruction. Half of those fired only transiently in responseto the instruction and the other half had dischargethat was sustained until the trigger signal or the beginningof movement. The majority of the neurons with sustained post-instructional activity discharged only in the “go” condition and not in the “no go” condition. The preparatory activitywas not related to eye movement occurring prior to limb movement. Schultz’s group also recorded neuronal activity in SMA in the same animals and found that relatively more SMA neurons were related to the instruction than were putamen neurons (Romo and Schultz, 1992). Just as movement-related activity in putamen is later than that in motor cortex and SMA, so set-related activity in putamen is later than that in cortex. In Alexander and Crutcher’s study, the average time of set-related activity in the putamen was 105msec after motor cortex and 230 msec after SMA (Alexanderand Crutcher, 1990b).Likewise,in Schultz’sstudy, the onset of the set-related discharge was later in putamen than in SMA (Romo and Schultz, 1992). Since set signals occur earlier in cerebral cortex than in striatum, it is likely that set signalsoriginate in cortex and are transmitted to the basal ganglia, rather than originating in the basal ganglia.

Although the late timing of movement-related activity argues against a role of the putamen in the initiation of movement, some neurons do have activity several hundred millisecondsin advance of movement.The activity of these neurons occurs after a movement-specific instruction but before the movement begins and is usually not time-locked to the movement. This type of neumnal activity is thought to be related to the animal’s “set” for the upcoming movement based on the preceding instruction. Tasks used to study preparatory activity generally have included the presentation of an instructional cue (“set”) followed after a variable delay by a signal to move (700msec after which a target stimulus was illuminated. The tasks were as follows:(l) ’’fixation” —themonkey wasrequired to maintain fixation and not saccade to the target; (2) “saccade” — the fixation point was extinguished when the target was illuminated and the monkey made a saccade to the target; (3) ’’saccadewith overlap’’-the target was illuminated while the fixation point was still on and themonkeywas requiredto saccadetothetargetonly after the fixation point was subsequently extinguished. This task was used to separate sensory responses to the target presentation from eye movement-relatedactivity; (4) ’’saccadewith gap”the fixation point was extinguished and 600msec later the target was illuminated. Because the target was the same over a block of trials, the monkey usually began the saccade before the target was illuminated. Therefore, these saccadeswere begunto a rememberedtarget location; (5) ’’delayedsaccade” —during fixation the target was briefly illuminated and then distinguished.After adelayof24 sec,the fixationpoint wasextinguishedandthemonkeymade a saccade to the remembered target location. The target was illuminated again 600msec after the fixationpointwasextinguished. Inthislattercase,the saccades were also to a rememberedtarget location, but the monkey was specificallyinstructed on the target locational the beginning ofeach trial. Saccade-related neurons were located in a longitudinal zone in the central part of thecaudate that receives input from the frontal and supplementary eye fields (Hikosaka et al., 1989a).Similar to limb movement-relatedputamen neurons, saccade-related caudate neurons increased activity above a low

393

baseline firing rate. Saccade-related neurons had broad movementfieldsmost ofwhich were locatedin the contralateral visual field. Many saccade-related neurons also responded to visual or auditory stimuli in the contralatcral field(Hikosaka et al., 1989b).In addition to saccade-related neurons with sensory responses, some caudate neurons had visual responses but no saccade-related activity. These visually-responsiveneurons had large receptivefields in the contralateral visualfield.Most of the responses were phasic ON responses (Hikosaka et al., 1989b). Other caudate neurons respondedto auditory stimuli in the experimental environment, such as keyboard sounds, the sound of the door opening, or footsteps. Previous studies also reported striatal neurons that responded to visual or auditory stimuli (Caan et al., 1984;Ro]ls et al., 1983). The time of onset of saccade-related caudate discharge varied across conditions with a mean of 49 msec prior to movement for saccades to visual targets and 106msecprior to movementfor saccades to remembered target locations (Hikosaka et al., 1989a).The timing of cortical activity in the frontal and supplementaryeyefieldswas not recordedin that study, but results from other studies suggest that saccade-related neuronal activity in these cortical areas appears to precede that in the caudate (Bruce and Goldberg, 1985;Schlag and Schlag-Rey, 1987). Just as someputamen neuronsdischargein relation to set in limb movementtasks, some caudate neurons have set-relateddischarge in eye-movementtasks. In Hikosaka’s study (Hikosaka et af., 1989a), 9% of caudate neurons fired in the “delayed saccade” task after the initial target presentation but before the fixationpoint was extinguished.The activityof nearly all of these neurons related to the direction of the upcoming eye movement and not to the location of the target. Thus, if the monkeymade a saccadein the wrongdirection,the preparatory dischargecorrelated with the direction of movement, not with the instructed direction. 3.3. Contextdependent Neuronal Activity in the Striatum

Several studies have reported striatal neuronal activity that appeared to be related only to movementsor sensorystimuli occurring in a specific context. Schneider and Lidsky (Lidsky et al., 1985; Schneiderand Lidsky, 1981)recorded neurons in the striatum of cats that fired during mouth movements if and onlyif the movementwas in relation to a tactile stimulus.Similarly,some neurons firedin responseto touch of the face only if that touch was in the setting of a facial movement. Hikosaka et al., 1989a)also described context-dependentactivity in the caudate. Some saccade-relatedneurons were active in relation to saccadesmade in one condition and not in others. 26V0of saccade-relatedcaudate neurons discharged oQlywhen the saccade was made to a remembered target location. Conversely, 21’7. discharged only when the saccade was made to a visual target. One-third of saccade-related caudate neurons were not context-specificand related to both memory-and visual-contingent saccades, Other caudate neurons had context-dependent responses to visual stimuli.

394

J. W. Mink

One type had a weak response to a visual stimulus outside of the task and a vigorous sensory response if a saccade was to be made to that stimulus (Hikosaka et a[., 1989b).Another type responded to a visual stimulus if it was used as the instruction cue for a subsequent memory-guidedsaccade. Another type of context-related activity has been seen in relation to stimulus-triggered (externally generated) or self-initiated (internally generated) movements.In the former, subjectsmovepromptly in response to a specificstimulus such as a light or the appearance a target. In the latter mode, subjects move at a time of their choosing. Schultz and his colleagues (Romo et al., 1992; Schultz and Romo, 1992)recorded neurons in the caudate and putamen of monkeystrained to reach for a reward in response to a light cue or at a time of their choosing. Of 39 neurons with activity during self-initiatedmovement, 29 also fired during externally triggered movements and 10fired only in the self-initiatedcondition. Of 53 neurons with activity prior to self-initiated movement, 27 were also active prior to externallytriggered movements and 26 were active only prior to the self-initiated movement. (Romo et al., 1992). The authors did not report how many neuronswereactive only in the externally-triggeredcondition. In the studies cited above, the authors took great care to assure that movementsperformed in different contexts were similar. However, they were not identical. For example, memory-guided saccades were slower and smaller in amplitude than were visually-guided saccades (Hikosaka et al., 1989a). Self-initiated arm movements were associated with somewhat less brisk onset of EMG activity and with more eye movements than were the stimulus-triggered movements (Schultz and Romo, 1992). Therefore, it could be said that the apparent context-dependent activity was due to differencesin the way the task was performed. Nonetheless, the apparent context-dependent neuron activity was robust and the movement differenceswere slight. Whilethe presenceof context-dependentactivityin the striatum has been emphasized, many striatal neurons are not context-specific.Furthermore, many different types of context-related activity have been described, suggesting that the striatum does not participate in movement in one context to the exclusion of all others. What these striatal neurons have in common is that their activity relates to movement. The numerous types of context-dependentactivity in the striatum are consistentwith the anatomy of the medium spiny cell and the corticostriatal projection. As described above (section 2.4.1), a medium spiny cell receives input from as many as 15,000cortical cells and each individual cortical input contributes few synapses. A medium spiny neurons requires temporal coincidence of many inputs in order to reach threshold. Furthermore, a medium spiny neuron may receiveinput from a number of cortical areas. Thus, a medium spiny neuron that receives input from somatosensoryand motor cortex may fire in relation to movements in the context of sensory stimuli (or vice versa). A neuron that receivesinput from frontal eye fields and prefrontal or parietal cortex may firein relation to saccadesto remembered

targets and one that receivesinput from frontal eye fieldsand visualareas may tire in relation to saccades to visual targets. There are severalimportant unansweredquestions about context-specificstriatal activity.What happens to context-dependent striatal activity when it is transmitted to the basal ganglia output nuclei? Do cells with different context-dependent activity convergeonto cellsin GPi and SNpr or are their outputs segregated?How does context-related activity in the striatum compare to that in the cortex? Context-dependent discharge has been described in premotor and supplementary motor areas (Crammond and Kalaska, 1994;Mauritz and Wise, 1986;Tanji, 1994), but it is not known whether any of the signals originate in the basal ganglia or if they originate in cortex and are transmitted to the basal ganglia.Based on the timing of movement-related and set-related activity in striatum as compared with cortical areas, it is likely that context-dependent activity does not originate in the striatum. Until these questions are answered, it can be said that context-dependent activity is feature of the striatum, but not necessarily the function. 3.4. Striatal Activity Related to Anticipation of “Behaviorally Significant” Events

Some neurons in both caudate and anterior putamen have been shown to discharge in apparent anticipation of predictableevents:In a go-no go task, striatal neurons discharged in relation to and preceding stimuli associated with the task (Apicella et aZ., 1992).These stimuli included the go-no go instruction, the trigger signal, and delivery of a reward. The associated neuronal activity was time-locked on the stimulus and could precede the stimulus by as much as 1.5 sec. Similarly,in the eye movement task of Hikosaka et al., 1989c),neuronal activity in the caudate that preceded and was time-locked to target appearance, fixation cue, or reward, In both these studies, the specific times at which stimuli o~urred were variable from trial to trial, but the sequenceof occurrencewas identical in each trial. In other words, the monkey could predict that an upcoming stimulus would occur, but not necessarily when it would occur. Other striatal neurons dischargedin relation to deliveryof reward, but after the reward was delivered (Apicella et al., 1991a;Hikosaka et al., 1989c).This activityappeared to be unrelated to any movement associated with consumption of the reward. The neuronal activitythat has been describedin the preceding sections has been thought to be that of medium spiny neurons because of the low discharge rates and narrow action potentials. Another type of striatal neuron has a tonic discharge rate of 2 to 10 spikes/secondand a broader action potential (Aosaki et al., 1994; Kimura et al., 1984, 1990). These tonicallyactiveneuronsare not activated by electrical stimulation of the globus pallidus (Kimura et al., 1990),are distributed throughout the striatum, and are thought to correspond to the cholinergic large aspiny interneurons (Aosaki et al., 1995). Their discharge bears no specific relation to movement (Apicellaet al., 1991b;Kimura et al., 1990),but they

TheBasalGanglia

395

do firein relation to certain sensorystimuli that elicit a subsequent movement (Kimura, 1986) or that precede reward (Aosaki et al., 1994;Apicella et al., 1991b). The stimulus-related neuronal discharge is dependent on the predictivefeatures of that stimulus. For example, a tonically active neuron may fire in relation to a clickingsound if the clickprecedesa fruit juice reward but will not fire to the click alone or to the reward alone (Aosaki et al., 1994).This type of neuronal responseis not necessarilymodality specific. If the tone predicting reward is substituted with a light that predicts the reward, tonically active neurons stop responding to a tone and begin responding to the light. This shift of responseoccurs gradually over a number of trials. If the reward is subsequently withheld on a series of trials, the response to the stimulus eventually extinguishes (Aosaki et al., 1994). What is the function of the anticipatory reward-related striatal activity? If the tonically active neurons are indeed cholinergic aspiny interneurons, they are anatomically situated to modify the activity of neighboring medium spiny cells. As discussed in above (section 2.4.4), one effect of cholinergicinput to medium spiny neurons is to stabilize their present state. Thus, in a situation where the current behavior results in reward, activity of the cholinergic interneurons would tend to reinforce the ongoing pattern of striatal activity. Although a small number of medium spiny neurons have predictive discharge, relatively few neurons in globus pallidus or SNpr have activity related to predictive stimuli or reward. (Brotchie et al., 1991c;Schultz, 1986).Thus, it would appear that the anticipatory and reward-related striatal discharge modifies the activity of striatal medium spiny neurons, but that most of these signals are not transmitted to basal ganglia output nuclei. Similar to the context-specificactivity, anticipatory and reward-related activity is a feature of the striatal processing, but is unlikely to be a function of the basal ganglia output.

proportion of STN neurons that respond to passive movements is unclear. Two studies have addressed this questionwith quite differentresults. In one study, only 20°/0of neurons that discharged in relation to active arm or leg movements also discharged in relation to passivemovements.(DeLong et al., 1985). In the other, ss~. discharged in relation to both active and passive movements (Wichmann et al., 1994a).Both studies found that the great majority of passiveresponseswere to joint rotation. However,in the former study, none of the tested cells responded to tendon taps, light touch, musclepalpation, or hair stimulation (DeLong et al., 1985).In the latter, 25°/0 responded in some degree to light touch (Wichmann et al., 1994a). It is not known to what degree STN neurons discharge in relation to movement parameters. The only study to address this question of amplitude or velocitycoding in STN included only 7 neurons, of which 3 had some correlation to amplitude and velocity(Georgopoulos et al., 1983).These numbers are too small to conclude that there is significant parameter coding in STN. Neuronal activity in relation to other variables has not been examined in the STN. It is not known if STN neurons have “set’’-related, anticipatory, or context-dependent activity in limb movement tasks such as is seen in striatum. The timingof STN activityis near the onset of limb movement,but is relativelylate compared with motor cortex and EMG. In one study, the median onset of activity change in STN was 50 msec before the start of a visuallyguided elbow movement(Georgopoulos et al., 1983).In that study, units were selectedbased on activity related to active movement. In another study, the mean onset of activitywas 2 msecafter the start of visuallyguided elbow movement(Wichmann et al., 1994a).In that study, the units were selected on the basis of activity in both active and passive movements. The timing discrepancy between those two studies may be due to selectioncriteria (neurons with sensory responses may fire later than those without) or small sample sizes. Alternatively, the 3.5. Subthalamic Nucleus Neurons are Tonically apparently different results may due to the way in Active and Somatotopically Organized which they were reported. In the former study, the Neurons in the subthalamic nucleus are tonically median time of activity change was reported and in active with average baselinedischarge rates of about the latter it was mean. If the distributions were 20 spikes/second(DeLong et al., 1985;Georgopoulos skewed,it is possiblethat the mean and median could et al., 1983;Matsumura et al., 1992;Wichmann et al., have differed by 52 msec. Although specific con1994a).They are organized somatotopicallywith the clusions regarding the absolute timing of STN lower extremityrepresenteddorsally and the faceand activity cannot be reached, the common finding eyes represented ventrally (DeLong et al., 1985; between the two studies is the relatively late timing Matsumura et al., 1992).Most subthalamic neurons compared with cortical motor areas. increasetheir dischargerate in relation to eyeor limb movement(Matsumura et al., 1992;Wichmann et al., 3.5.2. Subthalamic Nucleus Activity Related to 1994a). Saccades

The activity of neurons in the ventral portion of STN is related to saccadic eye movement. These neurons have been studied in tasks similar to one In monkeys trained to perform movements at the used in studies of caudate neurons (see Fig. 6) elbow, 60-75Y0of STN neurons had activity related (Matsumura et al., 1992). In those tasks, zs~. of to movement direction (Georgopoulos et al., 1983; task-related STN neurons were related to maintained Wichmann et al., 1994a). Of STN neurons that visual fixation, 23°/0 were related to saccades, 39°10 discharged in relation to limb movement, some also were related to visual fixation that was followed responded to passive movement. However, the immediatelyby reward, and 15°/0had visual sensory 3.5.1. Subthalamic Nucleus Activity Related to Limb Movement

396

J. W.Mink

responses.The great majority of task-related activity changes were increases. Most saccade-related STN neurons discharged in relation to saccades in the contralateral field. The saccade-related activity preceded the saccades, but often outlasted the saccades and often reached a maximum after the saccade had begun. 52Y0of the STN neurons had activity related to maintained eyeposition and not to saccades. Matsumura et al. (1992) interpreted this activity to be related to the suppressionof unwanted eye movements when fixation was required. This interpretation is consistent with the anatomy and physiology of the pathway from STN to SNpr to superior colliculus. Thus, increased activity in STN excitesSNpr whichin turn inhibits collicularneurons involved in saccade generation. This would tend to prevent saccades. The role of STN in suppressing unwanted movements will be discussed further in section 5. Some saccade-related neurons had greater increasesfor visually-triggeredsaccadesand others had greater increases for memory-triggered saccades (Matsumura et al., 1992).However,no STN neuron was exclusivelyrelated to either saccades to a visual target or saccades to a remembered target location. 3.6. Globus Pallidus Pars Interna and Subatantia Nigra Pars Reticulate are Somatopically Organized

compared to the activation of agonist muscles. In most studies the average onset of GPi activity was after the onset of EMG but before the onset of movement (Fig. 7) (Anderson and Horak, 1985; Brotchie et al., 1991b; Georgopoulos et al., 1983; Mink and Thach, 1991b;Mitchell et al., 1987).There was a tendency for GPi movement-relatedincreases to occur earlier than decreases(Andersonand Horak, 1985;Georgopoulos et al., 1983).The tendency of increases to occur earlier is consistent with the excitatory STN-GPi pathway being faster than the inhibitorystriatum-GPi pathway. One study reported that GPi activity precededagonist EMG on average, but it was by only 20 msec(Nambu et al., 1990).The late timingof GPi movement-relatedactivity suggests that the output of the basal ganglia is unlikely to initiate movement. Although movement-relatedactivity in GPi is late, some neurons have “set’’-related and cue-related activity that occurs earlier. These responses are not Cerebellum ● ✎ ❆ ✌✎✎

s. ,.. m ,,

~ ,m ---...

As noted in section’2.8, GPi and SNpr may be viewed as a functional unit that is separated anatomically by the internal capsule. The activity of , singleneurons in behaving animals tends to support .f ,.. this view. Neurons these structures have similar s !,. baseline firing rates. (DeLong et al., 1983;Hikosaka .mD and Wurtz, 1983d;Mink and Thach, 1991a;Schultz, : .1986) and there is a somatotopy with the body representedin GPi and the eyesand head represented in SNpr (DeLong et al., 1983).However, there may be some representation of limb movement in SNpr * (Schultz, 1986).

-,.. -!”--- -

-

-

-

@“

-

-

-

,-

:

‘--

-’ - -

-$ ,-

,-

(m..o1

T,”.

Agonist EMG

.”. -”. ..” -.--”----------

-

~: , - -.0 .-

, , , , , , , , , : : : : , 0 = - - - .- ,- -$ ,- - - - - -

GPi

!,-

.

3.6.1. Globus Pallidus Pars Interna Neurons are

,.-

s ,. . m .. . ,

m

Tonically Active During Maintained Posture and Change after Limb Movement is Initiated

Neurons in GPi firetonicallyat 60-80 spikes/seein monkeys maintaining a neutral limb position. (Georgopoulos et al., 1983;Mink and Thach, 1991a; Mitchell et al., 1987).Approximately 70?4.of arm movement-relatedGPi neurons increase activity and 30% decreaseactivity from this tonic baselineduring movement (Anderson and I-Iorak, 1985; Brotchie et al., 1991b;Georgopoulos et al., 1983; Mink and Thach, 1991b;Mitchellet al., 1987).The result of this pattern of activity change would be inhibition of a large part of thalamic and brainstem targets of GPi with disinhibition of a smaller part. It is tempting to speculate that this might form the basis for a center-surround organization with a facilitator center and an inhibitory surround (Mink and Thach, 1993).However, to date no study in normal animals has provided evidencefor anatomical segregationof GPi neurons that increase from those that decrease during movement. The timing of movement-relatedGPi activityis late

“

4

-

.0

,“

,“

,.

“.

ma.e.

“o

“e

-0

,,.. (....1 GPe

,

—

.‘ ‘. ,.-

s ● 0

T,F. (....1

Antagonist

EMG

● “ .

f’ s, . m *

LA -

.------- f!--!-?:::::::: :

i”.. --. ~-. ”-. ~ ---

----

---

.-, ---

.-, - --”!

-

Tmol ....)

Fig.7. Timingof GPiandGPeactivityin relationto onset of muscleandcerebelladentateactivityin a step-tracking wristmovementtask.Themovementbeginsat timeO(from Mink and Thach, 1991b); reprintedwith permissionof AmericanPhysiological Society).

TheBasalGanglia time locked to movement (Nambu et al., 1990; Neafsey et al., 1978). In a reaching task (Nambu et al., 1990), 160/0of GP neurons were cue-related, 2Lty0 were “set’’-related and 46?4.were movement-related and many of these neurons had more than one type of responses. In that study, GPe and GPi neurons werenot consideredseparately.Nambu et al. (1990) further tested each neuron by electrically stimulating prefrontal, premotor, supplementary motor, and motor cortex. Most GP neurons responded to stimulation of more than one area of cortex and, in general, most cue-related neurons responded to prefrontal and premotor stimulation, most “set’’-related neurons responded to prefrontal and premotor stimulation, and most movement-related neurons responded to premotor and motor stimulation. These resultsirylicatedthat at least some characteristics of the GPi output are determined by their indirect cortical inputs, but that there is convergenceof input from multiple areas of cortex through striatum and STN to GPi. Most of the responses to cortical stimulation consisted of an initial excitation followed by inhibition (Nambu et al., 1990),consistentwith a fast excitation through STN and a slower inhibition through striatum, as described in section 2.7. 3.6.2. Globus Pallidus Internal Segment Activity Does Not Corre[ate with Physical Parameters of Movement The correlation of discharge in GPi with specific parameters of movement has been investigated by two independentgroupswith similarresults(Brotchie et al., 1991b; Mink and Thach, 1991b). In both studies there was no consistent relationship between GPi activity and joint position, force production, movement amplitude or movement velocity during wrist movement. Although a previous study had reported a small number of GPi neurons with activity related to velocity and amplitude, in that task, velocity and amplitude were highly correlated and they may have been correlated with other variables, too (Georgopoulos et al., 1983). It has also been shown that few GPi neurons have activity correlated with the pattern of muscleactivity (Mink and Thach, 1991b; Mitchell et al., 1987). The lack of specific correlation of GPi activity with movement parameters suggests that the basal ganglia output is unlikely to control these parameters. Although parameter coding in GPi is weak generally, several studies have shown that a proportion of GPi movement-related neurons discharge in relation to movement direction. In studies of one-dimensional movements, the proportion of GPi units with directionaldischargevaried from 15Y0 to sf)~o. (Brotchieet al., 1991b;Georgopoulos et al., 1983;Mink and Thach, 1991b;Mitchell et al., 1987). When GPi units were compared across tasks, 41~0 correlated with the direction of movement in one wrist movement task, but only 25°/0 of those were also directional in a second wrist movement task (Mink and Thach, 1991b). That result suggeststhat GPi may not code direction per se, but instead may code some other variable that is correlated with direction in some tasks and not in others. JPN50+C

397

A better way to investigatedirectional signalsis to use a two- or three-dimensionalmovementtask. In a two-dimensionalarm movementtask, > 90°/0of GPi. neurons discharged in relation to direction (R.S.Turnerand M.E. Anderson,personal communication). This is a much higher percentage than has been reported previouslyand is probably due to the greater number of possible movement directions in the two-dimensionaltask. Neuronal signalsrelated to movementdirection are not unique to the basal ganglia and have also been describedin motor cortex, cerebellum,parietal cortex (see Georgopoulos (1994) for review). Thus, directional signalsare common in motor structures of the CNS. The presence of,directional signals in a given structure dpes not necessarily imply a role of that structure in the programming of direction. This is especially true in the basal ganglia where the directionalsignalsoccur after the movementhas been generated and hence cannot be used to program direction. 3.6.3. GPi Neuronal Activity is Not Exclusively Related to Any One Movement Mode or Context Basedon observationsof eyemovementsin people with Parkinson’s disease, Kornhuber hypothesized that the basal ganglia control slow “ramp” movements and not fast “ballistic” movements (Kornhuber, 1971). In an early experiment, GPi neurons were reported to discharge preferentially during ramp movements(DeLong and Strick, 1974). However,axial muscleswerealso more activein ramp movementsand it was unclear whether the GPi units were related to proximal muscles or to ramp movements specifically. Two recent studies have reexaminedthis question. One found that of 41 GPi units active during wrist movement, all were related to a ballistic movement and 80°Awere related to a ramp movement(Mink and Thach, 1991a).The other found that all of 11 GP (10 in GPe, 1 in GPi) units active in wrist movementswere active in both ramp and ballistic movements(Hamada et al., 1990).Both studies showed that the activity changes were somewhat greater in magnitude during ballistic movements than during ramp movements. These results are against exclusivecontrol of either ramp or ballistic movements by the basal ganglia. In addition to ramp and ballisticmovements,Mink and Thach, 1991a)compared visually guided ramp and sinusoidal movements to self-paced ramp and sinusoidal movements. Nearly four times as many GPi neurons were active in visually guided ramps than in self-paced ramps. However, a nearly equal number of GPi neurons were active in the visually guided and self-paced sine movements. No neuron was active exclusively in self-paced tasks. While individual neurons may have preferential activity in one movementmode, the overalloutput of GPi is not specific for any one of these movement modes. It should be further emphasized that a lack of phasic neuronal activitychangein a task does not mean that GPi does not participate in that task. The tonically maintained discharge in GPi neurons may play an important role by tonically inhibiting downstream movement generators to prevent them from becom-

398

J. W. Mink A

B

L t---

200

100

0

-..—------

/___

.J-----...— .—.—..—. L———.–

I I

Iillll .-.. ..-.— —-.:. .-, ---.-., ---- ...-, 100

illlltI

.— . -.. -. . -. L

0

, I

1 !

1 -.....4—-

...— .-...----1-----

‘--==’i%:7::zs:;: =——--——,=.—— .—.

.—

-—,

.

1

1

-

Fig. 8. Dischargeof a globus pa]lidus neuron (segmentnot identified)during a Sequential hold–movehold~mov~hol~ wristmovementtask. (A)‘Dischargewhenthe direction-of the‘second movementof the sequenceis predictable.(B)Dischargewhenthe secondmovementof the sequenceis variedrandomlyfromtrialto trial.Thetop graphsshowwristposition,themiddlegraphis a histogram oftheunitdischargeinspikes/s,thebottomgraphisa trialbytrialraster,Eachdivisiononthehorizontal scaleindicates1 s (fromBrotchiee?al., 1991c);reprintedwithpermissionof OxfordUniversityPress).

ing active during a task where they might otherwise interfere. Although relatively few GPi neurons were active in the self-pacedsinusoidal movement task, a lesion of GPi significantlyimpaired the performance of that task (Mink and Thach, 1991c). It has been suggestedthat some GPi neurons also have discharge patterns that are predictive of movement in the context of a previous movement (Brotchie et al., 1991c).In a hold-move-hold wrist movement task, Brotchie et al. found that the majority of GP activitychangedafter the onset of the movement, consistent with the timing reported in other studies (section 3.6.1). Some neurons also had a second burst of activity before the end of the final hold. To determine whether the second burst was related to the hold, expected reward, or termination of the trial, a second movement was added in sequence with the first. The task then consisted of hold–move-hold-move-hold. In that task, most neurons had two bursts, one followingthe onset of the first movementand the other followingthe onset of the second movement. One-third of GP neurons had an additional burst that preceded the second movement (Fig. 8). In all cases, this additional burst was much smallerthan bursts that followedthe onset of either movement.In 5 of 8 tested GP neurons, this additional burst correlated with the direction of the upcoming movement. Approximately half of GP neurons had greater activity when the second movement was predictable than when it was unpredictable. These neurons tended to be located in more anterior portions of GP, but it was not reported

whether they were in GPe or GPi. The authors suggestedthat this secondburst was predictiveof the second movement in the context of the first and that it may have played a role in the sequencing of movement (Brotchie et al., 1991c).They suggested that the first movement of any sequence is initiated by cortical mechanisms, but that the basal ganglia providethe signalsto initiate subsequentcomponents of the sequence. This idea had been proposed previouslyby Marsden (1987)and will be discussed more fully in section 5.3.3. A more recent test of the hypothesisthat the basal ganglia output is preferentially active in sequential movements has been performed by Strick and his colleagues(Mushiakeand Strick, 1995b;Strick et al., 1993).They trained monkeysto press buttons in two conditions. In the firstcondition, three of fivebuttons were illuminated in, a particular sequence. The monkey had to remember that sequence during a delay period and then press the buttons, one at a time, in the correct sequence.In the secondcondition, the monkey touched each button as soon as possible after it was illuminated.There was no requirementto remember the movement sequence in the second condition. One hundred fifty-five of two hundred thirty-sixGP neurons (the segmentwas not specified) had activity that was greater in one task condition than the other. Sixty-fivepercent of those were better related to the remembered sequence task. Of that 65’Yo,about half discharged in relation to a specific phase of the remembered sequence and one-third discharged in relation to a particular pattern of

TheBasalGanglia movements. Thus, about IOOAof task-related GP neurons had discharges that appeared to be dependent on the specific sequence that was to be performed. Those neurons were located in the dorsal-medialpart of GP which projects to premotor cortex via thalamus. Thus, a small percentage of GP neurons has activity that appears to be related to the preparation for movement in a particular sequence. A smaller proportion (4Yo) of neurons in the cerebella dentate nucleus has been reported to have sequence-selective discharge in the same task (Mushiake and Strick, 1995a).In a similar task, 8% of SMA neurons had similar sequence-selective discharge (Mushiake et al., 1990). Thus, activity related to movement sequences has been found in several areas of the motor system. The widely distributed activity related to sequential movements suggests that they are not a unique province of the basal ganglia, but that the basal ganglia may contribute to their performance. What do the basal ganglia contribute to sequential movements?It is likely that the contribution of the basal ganglia to sequential movements is similar to their contribution to movement generally. It is proposed below (section 5) that the basal ganglia focally select desired cortical and brainstem motor mechanisms and inhibit those that might compete with desired movement. At the time of transition from one element of a movement sequence to the next, the previously active motor mechanisms must be inhibited and the newly active mechanisms must be disinhibited. If there is an inability to focally inhibit competing mechanismsgenerally,it might be expectedthat the resultingmovementdeficitwouldbe compounded during a sequence of movements. This idea will be discussed further in section 5.

399

tralateral to the recording site. Half of the sensory responses were enhanced when a stimulus was presented in the context of a saccade. SNpr neurons that were active during saccades wereshownto project to the superiorcolliculuswhere they inhibited their targets (Hikosaka and Wurtz, 1983d).During saccades, the time of SNpr activity change was after that in the superior colliculus, a structure that is knownto be involvedin the initiation of saccades (Fig. 9) (Hikosaka and Wurtz, 1983d). Thus, for eyemovementsas wellas limb movements, the basal ganglia output acts after other structures initiate movement. The observation that the activity of all eye movement-related SNpr neurons decreases during movement and of a large majority of limb movement-related GPi neurons increases during movement is interesting and has been the source of considerable speculation (Hikosaka et al., 1993; Mink and Thach, 1991b). There are several possible explanations. First, the difference may result from fundamental differences between limb and eye movements.These differencesinclude the modalities of sensoryfeedback,number of degreesof freedomof movement,interaction of movementof one body part with another, and a greater requirement for postural stabilization of the body during limb movements. Decreased SNpr dischargeduring saccadesresults in disinhibition of collicular neurons to allow them to participate in the generation of saccades.By analogy, the decreased discharge of some GPi neurons would also remove inhibition from mechanismsinvolvedin the generation of limb movement.The increaseof the surrounding majority of GPi neurons would inhibit mechanismsinvolvedin the generation of competing postures or movements. Many of these competing mechanisms are reflexes that would potentially be triggered by the movement itself. However, in eye 3.6.4. Substantial Nigra Pars Reticulate Neurons movements,the movementitself may be lesslikelyto Discharge in Relation to Eye Movements activate ‘kompeting reflex mechanisms and thus Like the limb movement-related neurons in GPi, increased inhibition of other targets may be SNpr neurons that are related to saccadic eye unnecessary. The tonic discharge in SNpr during movements are tonically active (Hikosaka and visual fixation may be reinforced by increased STN Wurtz, 1983a). Unlike GPi neurons during limb activity during fixation and act to prevent unwanted movements, virtually all saccade-related SNpr neur- eye movementsby inhibiting the colliculus.A second ons have been reported to decreaseactivityduring the possibility is that normally the superior colliculus eye movement (Hikosaka and Wurtz, 1983a, 1983b, generates combined eye and head movements. In a normal state, an isolated eye movement would 1983c). In a series of experiment using tasks similar to require simultaneous inhibition of head movement those described in section 3.2., Hikosaka and Wurtz mechanisms. In the experiments of Hikosaka and (1983a, 1983b, 1983c)characterized the activity of Wurtz, the monkey’shead was held, thus eliminating saccade-related SNpr neurons. One-third of SNpr the need to suppressneck muscleactivity. It would be cellshad activityrelated to saccadesto a visualtarget. interestingto compare the dischargeof SNpr neurons None of the “saccade to visual target” neurons also during combinedeyehead orienting movement with changed activity during saccades to a remembered that during isolated saccadiceyemovements.A third target location or in relation to saccadesin the dark possibilityis that the differencesare due to sampling (Hikosaka and Wurtz, 1983a).Another third of SNpr bias, but this seems less likely because of the large cellshad activity related to saccadesto a remembered number of neurons that have been recorded. target location. (Hikosaka and Wurtz, 1983c).About half of the “saccade to rememberedtarget” neurons 3.7, Discharge of Globus Pallidus External Segment also had activity related to saccadesto a visualtarget, and Internal Segment are Similar but none were related to saccadesin the dark. Unlike Two types of neurons have been describedin GPe GPi neurons, a majority of SNpr neurons had sensory responses (Hikosaka and Wurtz, 1983a). based on their baseline activity patterns (DeLong, 58Y0of SNpr neurons had visual responsesand 9Y0 1971).Most fire at a high frequency(70 spikes/sechad auditory responses with receptive fields con- ond) that is interrupted with long pauses. A smaller

J. W. Mink

400

number fire at a low frequency (10 spikes/see on average) and have frequent spontaneous bursts of activity. Although the baseline discharge patterns in GPe differfrom GPi, the activity during movementis remarkably similar in the two structures. GPe neurons change activityin relation to limb movement and for the majority, these changesare an increasein activity (Anderson and Horak, 1985;DeLong, 1971; Mink and Thach, 1991b;Mitchell et al., 1987).As has been described for GPi, the coding of movement amplitude and velocity and of muscle length and force is weak (Brotchie et al., 1991b; Mink and Thach, 1991b). Individual GPe neurons may have a preferential relation to one movement mode or movementin a particular context, but overallthere is

no exclusiveactivity of GPe in relation to one mode or context. (Brotchie et al., 1991b;Mink and Thach, 1991a).The timing of movement-related activity in GPe is late (Georgopoulos et al., 1983; Mink and Thach, 1991b). The similarity between GPe and GPi activity probably reflects the similarity of their inputs. However,their outputs are quite different:GPe sends inhibitory projections to STN and GPi. It is somewhat surprising that the majority of neurons in both GPe and GPi increase since GPe inhibits GPi both directly and indirectly through STN. The similarity of discharge suggests that GPe activity modulates or focuses GPi activity rather thap being a primary determinant of GPi activity. The specific

A

H

.:.;,:t;::.;~--+ 1 -----------. :---~---~+-; : ..... . —.. --.— ----------.— +....— H v

1

. . . .. .”-” ... ..

-

....... .—— ..... ... . ..--.. ::--:”:::.;-:.:~i--::::”:” ..-.. ............. .... . .. . ..-.-.—--------... . . -------.... .. ..-..-.,-- ---. .... -“-.-..-:-?. ------

.. -“.------—. . .. . .,., ------------

-=7==’:: =-. .:”:?.: ”.- “-””--”--..-.”.-’.-”-““.” :--:--:::;b:;b-. . .. ..... ..“’.’..-.— =’=---=:=::::: —--------............... ---1-. .....

1

.........——.—

P ;!

.“.

-

I

,,,

.,..

,,

2

[-_.d4 .,,,,,,,,

007-030

. .

.......... .... ........ . . .::-.:.-.:..:.-..:.+:::::”..: ,..-.--—.... -....— --------..- - -...-. ...... .-—-----..

007-029 “ ‘

,,

.

-..-......,.—.— .’ .-....-...— ;,:::::;:,: -~+,::-.::.-. .. .-,.,:,: .-.. . ..

,, Ao...do,a 1

,,,.

[>

..... ! ,.,

,, . ---

--.,,

,,,

,,,

,,

200 MSEC

Fig.9. Dischargeof SNprunitand superiorcolliculusunitduringsaccades.(A)Dischargerelatedto a visualsaccadeto target(saccadewithoverlaptask(Figure6)).(1)SNprneuronwithdecreaseddischarge; (2)superiorcolliculusneuronwithincreaseddischarge.NotethelatetimingoftheSNprdecreaserelative to thesaccadeandcollicularincrease.(B)Dischargeof thesameneuronrelatedto a memory-contingent saccade(delayedsaccadetask (Figure6)).Abbreviations— F: fixationtarget;T: saccadetarget;H: horizontaleye position;V: verticaleye position(fromHikosakaand Wurtz, 1983d);reprintedwith permissionof AmericanPhysiological Society).

TheBasalGanglia

No task -1.5

-1.0

-0.5

1 Light

0.5

401

A

/ GPi) slowed elbow movements in a tracking task (Here et al., 1977;Here and Vilis, 1980).The slowedmovementwas accompanied by cocontraction of agonist and antagonist muscles with a flexor bias. Cooling of GP impaired movements performed either with or without visual feedback. Similar results were found by Mink and Thach, 1991c) in one monkey after muscimol injections or a kainic acid lesion involving both pallidal segments.In that study, the monkeyhad slow movement, agonist and antagonist cocontraction with a flexorbias, and impairment in tasks with and tasks without visual feedback. Horak and Anderson (1984)showed that kainic acid lesions of the globus pallidus(GPe > GPi) slowedmovementin a reaching

J. W. Mink

408

previously(Mink and Thach, 1991a):visually-guided step-tracking, visually-guided ramp-tracking, visually-guided sinusoid-tracking, and self-paced sinusoidal movements. After inactivation or ablation, movement was slowed in all conditions but reaction time was normal. Both visually-guidedand self-paced movementswere impaired and movementsoutside of the tasks were slow and accompanied by abnormal limb postures. The tasks were performed with abnormal tonic and phasic cocontraction of agonist and antagonist muscleswith a slight flexor bias. By using torque loads, Mink and Thach were able to compare the effectof GPi lesionon movementsmade by turning on agonist muscles with those made by turning off the antagonists. It was shown that GPi lesionscaused greater impairment when the monkey had to turn offan active antagonist musclethan when it had to turn on an active agonist (Fig. 11). The cocontraction and inability to turn off the active antagonist during movement suggestedthat the GPi acts to prevent unwanted muscle activity from interfering with voluntary movement (Mink and Thach, 1991c,1993). Several other studies have shown similar movement abnormalities after lesions restricted to GPi. The first description of a kainic acid lesion restricted to GPi in the monkey reported slowing of elbow

task. In that study, there was no apparent agonist–antagonist cocontraction. On the contrary, task-related EMG was decreased and the authors attributed this to a decreased ability to scale the activity of agonist muscles. The reason for the different EMG abnormalities across these studies is not known. They may have been due to differencesin lesion location or task requirements. Despite the different levelsof muscleactivity, all of these studies found slowing of movement but no prolongation of reaction time after GP lesions, suggesting that movement initiation mechanisms remain intact after these combined GPe and GPi lesions. 4.2.3.2. Experimental lesions restricted to GPi Lesions restricted to GPi result in slowness of movement of the contralateral limbs with abnormal muscle activity. Mink and Thach, 1991c) made injections of muscimol and kainic acid that were restricted to GPi in one monkey. The effects of temporary inactivation with muscimol and of permanent ablation with kainic acid were qualitatively similar but the deficit was greater with the permanent ablation. They studied the effect of GPi inactivation and ablation on wrist movementsin four tasks in which singleunit activity had been recorded

ExtensorsLoaded

FlexorsLoaded

l---r b!!c5~~k A

I

B

50”

Flexion

-1200

-600 t

O

600

1200

c

-1200 -600 t

I

I

600

1200

D —

Extension *

d 1

-1200 -600 t O

J #

I

600

1200

1

-12C0 -600 t

O

1

t

600

1200

Fig.1I. Slowingof wristmovementafterGPilesion.(A)and (D)showmovementsmadebyturningon the loadedmuscleto makethe movement(flexionwithflexorsloadedand extensionwithextensors loaded).(B)and (C) showmovementsmadeby turningoffthe loadedmuscleto makethe movement (flexionwithextensorsloadedand extensionwithflexorsloaded).Theuppertracein eachgraphshows wristposition,thelowertraceshowswristvelocity.Thesolidtracesshowcontroltrials,thedottedtraces showtrials 24 days after a kainicacid lesionof GPi (fromMinkand Thach, 1991c);reprintedwith permissionof AmericanPhysiological Society).

TheBasalGanglia movementin a tracking task and an associatedflexor drift during attempts to maintain elbow position (DeLong and Coyle, 1979).Muscimolinjectionsinto GPi in monkeys,performing a two dimensional arm movement caused slowing of movement with cocontraction and a flexor bias (Anderson et al., 1993). Electrolytic lesions of GPi caused increased EMG and increased resistance to passive stretch at the elbow (Rea and Ebner, 1991). Cooling or electrolytic lesion of GPi in monkeys performing a reaching task slowed movement and decreased its accuracy (Trouche et al., 1979).Kato and Kimura (1992)reported decreased accuracy of elbow movements with overshoot or undershoot after injectionof muscimol or kynurenate into one GPi site in each of three monkeys. Some of their injections also caused increased antagonist muscle activity and slowing of movement. However,other of their injectionshad no effecton average movement speed or amplitude. The lack of effectof some of their injectionson movement speed may have been due to the fact that their task was performed without torque loads. The lack of effectof some of their injections on movement speed may have been due to the location of the injection site. The site that produced agonist–antagonist cocontraction in Kato and Kimura’s(1992)study was in an area of the ventromedialGPi that corresponded to the site of the GPi lesion in Mink and Thach’s (1991c)study. Reaction time is normal after GPi lesions (Kato and Kimura, 1992;Mink and Thach, 1991c),or may even be decreased(Amato et al., 1978;Trouche et al., 1984).The fact that reaction time is not increased after ablation of GPi, the basal ganglia output for limb movements, suggests that this circuit is not critically involved in the initiation of movement. As discussed above, it has been suggested that reduced activity in GPi is the underlyingmechanism of chorea and hemiballismus that is seen after ablation of the excitatory input from STN (Albin et al., 1989; Crossman, 1987; DeLong, 1990). However, chorea or hemiballismus has not been reported after experimental ablation or inactivation of GPi. Injection of kynurenate (an excitatory amino acid antagonist) into GPi has been reported to produce involuntary movementsthat were said to be indistinguishable from those seen after STN lesions (Robertson et al., 1989).This would tend to confirm the hypothesis that dyskinesia after STN lesions is due to the loss of excitatory input to GPi. However, Kato and Kimura (1992) were unable to produce dyskinesiawith kynurenate injections.This may have been due to dose differences or differences in the location of the injection sites. Although chorea and hemiballismusmaybe due to diminishedexcitationof GPi by STN, elimination of GPi activity does not produce these movements. One way to explain these findingsis to postulate that partial reduction of GPi activity causes target neurons in the thalamus and brainstem to become unstable and fire in bursts, ultimately leading to random bursts of muscle activity. Elimination of GPi activity results in a more stable state of the thalamic and brainstem target neurons, but at a higher firing rate that would lead to sustained activityin the musclesand cocontraction rigidity.

409

4.2.3.3. Efect of GPi lesions on the activity of downstream targets To understand the mechanism of movement abnormalities that result from focal lesions, it is important to know how the change in behavior relates to changed activity in structures downstream from the lesion. Anderson et al. (1993) recorded the activity of singleneurons in the part of thalamus that receivespallidal input in monkeys before and after inactivation of GPi with muscimol. As noted above, injection of muscimol into GPi resulted in a drift of the arm during attempts to maintain a position. Accompanyingthe drift was an increase in the tonic discharge rate of thalamic neurons and muscular cocontraction. After muscimol injection in GPi, the timing of the phasic movement-relateddischarge of thalamic neurons was unchanged. The magnitude of phasic movement-related thalamic discharge was increasedfor someneurons,decreasedfor others, and unchanged for still others. These results confirm that GPi exerts a tonic inhibitory effecton thalamus that increases for some neurons and decreasesfor others during limb movement,but that the timing of phasic discharge in thalamus is determined by other, presumably cortical inputs. 4.2.4. Experimental Lesions of Substantial Nigra Pars Reticulate Impair Eye Movements

Due to its proximity to and interdigitation with SNpc, it has been impossibleto produce electrolytic lesions restricted to SNpr. However, with the use of microinjectionsof the GABA agonist muscimol it is has been possible to inactivate neurons focally in SNpr (Hikosaka and Wurtz, 1985b).Hikosaka and Wurtz, 1985b) injected muscimol into the lateral SNpr where they had previously recorded neurons that were related to saccadiceye movements(section 3.6.4). Inactivation in this area resulted in involuntary saccades and an inability to maintain visual fixation (Fig. 12). Voluntary saccades were affected only slightlyby inactivation of SNpr. Saccadesto the contralateral field were slightly increased in amplitude and saccadesto the ipsilateral fieldwere slightly decreasedin amplitude. Saccadesto the contralateral field had reduced latency (reaction time) while those to the ipsilateralfieldhad increasedlatency. Saccades to rememberedtargets were more affectedthan were saccades to visual targets. When the monkeys made spontaneous eye movementsafter SNpr inactivation, the eyes tended to assume an abnormal position in the contralateral field. The inability to suppress involuntary saccades after inactivation of SNpr appears to have resulted from disinhibition of the superiorcolliculusbecause(1) injectionof the GABA antagonist bicuculline into the superior colliculus mimickedthe effectsof muscimolin SNpr (Hikosaka and Wurtz, 1985a)and (2) superiorcolliculusneurons fired at a higher rate after SNpr inactivation (Hikosaka and Wurtz, 1985b). Thus, just as GPi inactivation results in abnormal excess limb and trunk muscle activity, SNpr inactivation results in abnormal excesseye movements. There are differences between the effect of GPi inactivation on limb movement and the effect of

J. W.Mink

410

A

10 DEG

111111111

I

111111111

100 MSEC

r

1

L

J

B

j224n

—. Fig.12.SaccadicjerksafterinactivationofSNprwithmuscimol.(A)Superimposed tracesofeyeposition duringattemptsto maintainfixations.Uppertracesshowhorizontaleyeposition;lowertracesshow verticaleyeposition.(B)Trajectoriesof”saeeadic jerksduringattemptsto maintainfixationat thecenter point(fromHikosakaandWurtz,1985b);reprintedwithpermissionof AmericanPhysiological Society). SNpr inactivation on eye movement that are consistent with the difference in movement-related discharge patterns in the two structures. Recall that nearly all SNpr neurons with saccade-relatedactivity decreased firingduring saccadeswhilethe majority of

GPi neurons with limb movement-related activity increased. Since the output of these structures is inhibitory, inactivation results in disinhibition of their targets. In SNpr, where neurons are tonically active during fixation and decrease with saccades,

The Basal Ganglia

muscimolproduces a state similarto that seenduring saccades and the result is involuntary saccades but relatively normal voluntary saccades. This occurs because the saccade generators in superior colliculus and frontal eye fields are activated normally for voluntary saccades but are inappropriately active during attempted fixation.In GPi, muscimolnot only causes disinhibition of downstream mechanismsthat are normally disinhibited in the desired movement, but it would also disinhibit competing mechanisms. Since the limb movement generators in cerebral cortex are intact, limb movement is initiated normally. However, movement generators for other motor patterns are tonically disinhibited leading to muscular cocontraction that interferes with the desired movement.

4.2.5. Experimental Lesions Restricted to Globus Pallidus Pars Externa

There have been fewreports of lesionsrestricted to GPe. Injection of muscimol into GPe at one site in one monkey resulted in abnormally flexedposture of the arm with agonist–antagonistcocontraction (Kato and Kimura, 1992). Blockade of excitatory transmission in GPe of monkeys with kynurenate (Kato and Kimura, 1992)also caused muscular cocontraction. These results are similar to those of Here and Vilis(1980)whosecooling probes were closer to GPe than to GPi, but differ from those of Horak and Anderson (1984)whose lesionsalso involvedGPe to a greater extent than GPi. The results of Kato and Kimura’s GPe inactivation are similar to those of several groups that inactivated or ablated GPi (Anderson et al., 1993; DeLong and Coyle, 1979; Mink and Thach, 1991c). Injection of the bicuculline into GPe produces chorea that is similar to that seen with STN lesions (Matsumura et al., 1995;Mitchell et al., 1989b).It has been thought that the mechanism of the chorea after bicuculline injection in GPe is disinhibition of GPe which leads to inhibition of STN. Thus, when chorea is produced by injectingbicucullineinto GPe, 2-DG activity is increased in STN and decreased in GPi (Mitchell et al., 1989b).Matsumura et al. (1995) tested this hypothesis by recording the activity of single neurons in GPe and GPi after injecting bicuculline into GPe. As was expected, they found that most GPe neurons increased activity after injectionof the GABA antagonist. Unexpectedly,the majority of GPi neurons also increased activity after the bicucullineinjection. In addition, many GPe and GPi neurons developed abnormal bursts and pauses in association with the dyskinesia. Thus it appears that chorea is associatedwith an abnormal pattern of discharge with increases in some GPi neurons and decreases in others. Based on a small sample, there were some areas of decreasedactivity surrounded by areas of increased activity in GPi, suggestive of a center-surround pattern. These results support the idea that chorea is due to abnormal patterns of activity in GPi and not just globally decreased GPi output.

411

4.2.6. Experimental Pars

Compacta

Lesions of Substantial Nigra Mimic Parkinson’s Disease

As noted above, the interdigitation of SNpc and SNpr have made selectiveelectrolyticlesionsof either nucleus difficult to produce. One way around this problem has been to use 6-hydroxydopamine (6-OHDA) to produce a specificlesion of catecholamine neurons (Ungerstedt, 1971). Injection of 6-OHDA into the substantial nigra bilaterally produces selectiveloss of SNpc dopamine neurons. Lesions made in this way result in some of the abnormalities of Parkinson’s disease, namely the slownessand paucity of movement and rigidity, but tremor is rare (seeDeLong and Georgopoulos(1981) for review). In 1982, a contaminant of a synthetic heroin analog was found to produce parkinsonismin those who injected it (Langston et al., 1983).The contaminant was identified as l-methyl-4-phenyll,2,5,6-tetrahydropyridine (MPTP). MPTP is oxidized in the brain by monoamine oxidase to MPP+ which is taken up by dopamine neurons where it inhibits oxidative metabolism in the mitochondria and ultimatelyleads to cell death (Tipton and Singer, 1993). Subsequently it has been found that MPTP produces a parkinsonian syndromein severalspecies of monkeys(Benazzouzet al., 1992;Crossman et al., 1987; Jenner et al., 1984; Schneider et al., 1988; Schultz et al., 1985, 1989). Unlike monkeys given 6-OHDA, monkeys given MPTP had tremor in addition to slowness of movement, paucity of movement,and rigidity.The parkinsonian syndrome in monkeys given MPTP has been associated with nearly complete degeneration of dopamine neurons in the SNpc and adjacent ventral tegmental area with variable degeneration in the locus coeruleus (Crossman et al., 1987).It may be involvementof the latter two structures that make the MPTP monkey a more complete model of Parkinson’s disease. MPTP monkeys improve when given the dopamine precursor L-DOPA, and have side effects of chorea when they are given too much L-DOPA, similar to what happens in people with Parkinson’s disease (Crossman et al., 1987).Hence, by behavioral, pharmacological, and pathological measures, the MPTP monkey is an excellentmodel of human Parkinson’s disease. When monkeys were given MPTP via unilateral intracarotid injection, they developed behavioral abnormalities within 2 to 4 days (Benazzouz et al., 1992). Initially they displayed preferential turning toward the lesioned side followed by a progressive rigidity with flexedposture and decreasedmovement of the contralateral limbs. In a two-dimensionalarm movement, monkeys with unilateral MPTP parkinsonism had prolonged reaction time, decreased peak velocity, and reduced movement amplitude (Camarata et al., 1992).The movementdeficitswere greater for larger amplitude movementsand for movements that weredirected away from, rather than toward the body. The deficits all improved after the monkeys were given L-DOPA,a dopamine precursor. Monkeys that were given MPTP intravenously or intraperitoneally developed a severe bilateral syndrome with flexed posture, rigidity, and lack of

412

J. W. Mink

spontaneous movement (Jenner et al., 1984; uptake in a structure reflects the activity of the Schneider et al., 1.988;Schultz et al., 1985).They afferentsto that structure. Based on this, the pattern often did not eat or drink spontaneously for several of 2-DG uptake in the MPTP monkey was weeks and required either supplemental feeding or interpreted as reflectingincreased striatal inhibition treatment with L-DOPA to reverse the symptoms of GPe with resulting disinhibition of STN and enough that they could feed themselves.Eventually excessiveactivation of GPi (Crossman, 1987).This they were able to eat and drink without treatment, schemeis supported by findingsof increasedlevelsof but the other motor abnormalities remained. The met-enkephalinimmunoreactivityin the striatum and severe motor deficits were associated with marked increased met-enkephalin mRNA levels in GPe of loss of dopamine cells in SNpc and of dopaminergic MPTP monkeys (Augood et al., 1989) and in terminals in the striatum. 6-OHDA treated rats (Gerfen et al., 1990). It has Monkeyswith MPTP parkinsonismhad both S1OW been suggested that increased striatal inhibition of movement and prolonged reaction times (Schultz GPe and decreased striatal inhibition of GPi are et al., 1985, 1989).The slowing of movement was mediated by the differential effect of dopamine on accompanied by cocontraction of agonist and these two pathways: inhibition of the striatal-GPe antagonist muscles(Fig. 13)(Benazzouzet al., 1992). pathway via D2 receptors and facilitation of the The increased reaction time was accompanied by an striatal-GPi pathway via D1 receptors (see section increasein latency from the cue to move to the onset 2.4.2). of EMG activity showingthat the deficitwas not just Electrophysiologicalevidence has tended to supone of muscle contraction rate. The reaction times port the idea that parkinsonism is associated with tended to recover over time, but the slowness of increased activity in GPe and decreased activity in movement did not improve (Schultz et al., 1989).It STN and GPi. DeLong and colleaguesrecorded the is interesting that in a block of more than 100trials, activity of single neurons in the GP and STN of the reaction times were often normal in the first 20 monkeyswith MPTP parkinsonism (Bergman et al., trials, but became more prolonged as the number of 1994;Miller and DeLong, 1988;Wichmann et al., trials increased. It has also,been noted that monkeys 1994b).The average discharge rate of STN neurons with MPTP parkinsonism often stopped performing was slightly increased in parkinsonian monkeys a task, but would perform again if they were guided (Bergmanet al., 1994).This was particularly true for through the task or are vigorously stimulated in a subpopulation of STN neurons that developedan nonspecific ways (Schneider et al., 1988; Schultz abnormal 4-8 Hz oscillatory discharge after MPTP. Millerand DeLong(1988)showedin MPTP monkeys et al., 1985).These findings suggest that fatigue or decreased motivation may contribute to the deficits that the average baseline discharge rate increased in GPi and decreased in GPe. Similar results were seen in the MPTP monkey. When monkeys with MPTP parkinsonism were obtained by Filion and Tremblay (1991), but given L-DOPAthe reaction times, movement speed, Bergman et al., (1994) found no change in the and rigidityimproved(Schneideret al., 1988;Schultz average discharge rate of GPi neurons after MPTP. 1989). It is interesting that despite the GPe and GPi neurons also had abnormally increased et al., improvement of these measures of movement with responses to somatosensory stimuli, passive limb L-DOPA, the monkeys continued to have long movement, and electrical stimulation of the striatum periods between trials and would often quit (Bergman et al., 1994;Tremblay et al., 1989).These performing the task for long periods of time increased responses were also less specific than, in (Schneider et al., 1988).This suggeststhe possibility normal monkeys. A further test of the hypothesisthat parkinsonism that non-dopamine mechanisms may be responsible for some aspects of the decreased task performance is associated with decreased activity of GPe and increased activity of GPi and STN comes with focal in the MPTP monkey. The MPTP model of Parkinson’sdiseasehas been lesions of the presumed overactive structures. useful for determining changes in the physiologyof Ablation or inactivationof STN has been reported to downstream structures that are associated with the reverse the symptoms of MPTP parkinsonism parkinsonism. One approach has been to use 2-DG (Bergman et al., 1990; Wichmann et al., 1994b). autoradiography. Another approach has been to When STN was inactivated, transient dyskinesiawas measure directly the content of neurotransmitters or seen initially but when it resolvedthe monkeys were their mRNA messagein synaptic terminals. Finally, said to move normally. Similarly, blockade of the someinvestigatorshave recordedthe activityof single excitatory input from STN to GPi with kynurenate neurons in downstream structures after MPTP reversed parkinsonism in the MPTP monkey administration. The effect of 6-OHDA lesions of (Brotchie et al., 1991a).No quantitative measures of SNpc on the activity of downstream structures is movement after STN inactivation in the MPTP similar to that of MPTP. These results have been monkey have been reported. It is interesting that reviewed extensively elsewhere (Albin et al., 1989; symptomatic improvement after STN inactivation Crossman, 1987;DeLong, 1990;Gerfen et al., 1990). was not accompanied by normalization of baseline In MPTP monkeys, there was increased 2-DG discharge in GPi (Wichmann et al., 1994b). This activityin GPe, the pedunculopontinearea (midbrain suggeststhat the abnormal tonic discharge in GPi is extrapyramidal area), and VA/VL thalamus with not the sole etiology of the symptomsof Parkinson’s slightly increased activity in GPi and markedly disease. Lesioning of GPi is an old approach to the reducedactivityin STN (Mitchell et al., 1989a).Since 2-DG uptake is thought to correlate with the activity treatment of Parkinson’s disease in people (Cooper of nerve terminals (see Crossman (1987)for review), and Bravo, 1958). The so-called pallidotomy was

TheBasalGanglia

413

largely abandoned when L-DOPAwas introduced in above, pallidotomy has been reintroduced using the 1960s as a treatment for Parkinson’s disease. modern stereotactic and physiological techniques Recently, in large part because of the data described (Baron et al., 1993; Goetz et al., 1993).The early

MPTP- TREATED

NORMAL FLEXION MECHANOGRAM 200ms E.M.G AITIVIT

I mvL 200ms

‘

Antagonist

muscle

EXTENSION MECHANOGRAM —. .. ... -

200ms E.M.G ACTIV

1 mvL 200ms

Ant agonist

Fig. 13. Impaired movementand cocontraction in the MPTP monkey. Mechanogramrepresentation of elbow position and agonist and antagonist muscleEMG in a tracking task before and after,MPTP (from Benazzouzet al. (1992);reprinted with permissionof Springer-Verlag).

414

J. W. Mink

reports of the modern pallidotomy suggestedthat it can be quite effectivein alleviating the symptoms of Parkinson’s disease (Baron et al., 1993; Laitinen et al., 1992).These results also tended to support the model described above. Surprisingly, the most consistent beneficialeffectof pallidotomy may be the reduction of dyskinesiasthat are inducedby L-DOPA treatment (Laitinen et al., 1992).However,L-DOPAinduced dyskinesia, like chorea and hemiballism, is thought to be due to decreased activity of STN and GPi, not to increased activity (Albin et al., 1989; Crossman, 1987;DeLong, 1990).Thus, the situation with pallidotomy and L-DOPA-induceddyskinesia may be similar to that with hemiballism and GPi lesions: partial reduction or possibly bursting of activity in GPi causes involuntary movements and elimination of that activity eliminates these movements (see section 4.2.2). Abnormally increased GPi activity should lead to excessiveinhibition of thalamic targets and reduced activity in motor cortical areas. Although single neurons have not been recorded in the thalamus of parkinsonian monkeys, the activity of motor cortex neurons has been recorded. In monkeys that have been given 6-OHDA or MPTP, the activity of motor cortex neurons during arm movementsis abnormal in severalways (Doudet et al., 1990;Gross et al., 1983; Watts and Mandir, 1992).First, the peak activity is reduced in movement-relatedneurons in the parkinsonian monkey. Second,the timing of the movementrelated changes is more variable, but there is not a consistent delay in the onset of motor cortex activity. Third, the rate of increase of movement-related activity is reduced. These results are consistent with excessive inhibition from GPi to thalamus, which would be reflectedin decreased excitabilityof motor cortex. However, there is also a dopamine input to motor cortex (Gaspar et al., 1992),and the abnormal motor cortex dischargemay be due to the loss of that input in addition to the abnormal GPi discharge.This issue could be resolved by injecting 6-OHDA into motor cortex to locally destroy dopamine terminals and comparing the effect on motor cortex activity with that seen in the MPTP monkey. Overall, the evidence would seem to support the model that excessive GPi discharge is associated with parkinsonism and may be the primary underlying physiological mechanism. However, the facts that average baseline discharge of GPi neurons in some parkinsonian monkeys is normal and that reversal of parkinsonism by STN is not accompanied by normalization of GPi activity suggest that the model is incomplete. One shortcoming of the model is its emphasis of tonic neuronal discharge rates, when it is known that basal ganglia structures have prominent phasic discharge patterns during movement. More important may be the abnormal temporal patterns of neuronal discharge that have been reported in the MPTP monkey (Bergman et al., 1994; Tremblay et al., 1989). Another shortcoming of this model is that the direct effect of dopamine on STN and GPi is not taken into account. Although small by comparison with the nigrostriatal dopamine projection, these dopamine projections may be important in the pathophysiology of parkinsonism.

5. FOCUSED SELECTION AND INHIBITION OF COMPETING MOTOR PROGRAMS: AN HYPOTHESIS OF BASAL GANGLIA FUNCTION

The results of anatomical, physiologicaland lesion studiescan be integratedinto an overallhypothesisof what the basal ganglia contribute to movement.This hypothesis states that the tonically active inhibitory output of the basal ganglia acts as a “brake” on motor pattern generators (MPGs) in the cerebral cortex (via thalamus) and brainstem. When a movement is initiated by a particular MPG, basal ganglia output neurons projecting to competing MPGs increase their firing rate, thereby increasing inhibition and applying a “brake” on those generators. Other basal ganglia output neurons projecting to the generators involved in the desired movementdecreasetheir discharge,thereby removing tonic inhibition and releasingthe “brake” from those desired generator. Thus selected movements are enabled and competing postures and movementsare prevented from interfering with the one selected. How might this mechanismwork (Fig. 14)?When one makes a voluntary limb movement, that movement is initiated by mechanisms in prefrontal, premotor, supplementarymotor and primary motor cortex and the cerebellum.Premotor, supplementary motor, and primary motor cortex send a corollary signalto STN, excitingit. STN projectsin turn to GPi in a widespread pattern and excites GPi. This increased GPi activity causes inhibition of thalamocortical and brainstem motor mechanisms.In parallel to the pathway through STN, signals are sent from all areas of cerebral cortex to striatum. The cortical inputs are transformed by the striatal integrative circuitry to a focused,context-dependentoutput that inhibits specific neurons in GPi. The inhibitory :triatal input to GPi is slower, but more powerful, than the excitatory STN input. The resultant focally decreased activity in GPi selectivelydisinhibits the desired thalamocortical and brainstem MPGs. Indirect pathways from striatum to GPi (striatum+GPe+GPi and striatum+GPe+STN+ GPi) result in further focusingof the output. The net result of basal ganglia activity during a voluntary movement is the braking of competing motor patterns and focused release of the brake from the selected voluntary movement pattern generators. 5.1. Why Must Competing Motor Pattern Generators be Inhibited During a Movement A large number and variety of MPGs must gain accessto the motor neuron pools in order to produce posture and movement. Prior to the final common pathway, the various MPGs may act through common descending pathways and interneurons in the brainstem and spinal cord. Activation of these pathways by one ,MPG may produce an action that is in direct competition with the action of another. Simultaneous activation of,competing MPGs would result in ineffectiveaction and cause inappropriate muscular cocontraction and abnormal postures and movements,Consider again the example of reaching to pick an apple from a tree (section 1). If the output

The Basal Ganglia

Striatum

~ M

GPi

~ ~

‘

~

STN

VLo\MEA

Other Motor Programs

v

Other Motor Programs

DesiredMotor Program

Emits@’

S

Fig. 14. Relationship of GPi activity to inputs from striatum and subthalamic nucleus. During voluntary movement, excitatory subthalamopallidal neurons increase the activity of the pallidal neurons in the territory surrounding a functional center. Inhibitory striatopallidal neurons inhibit the functional center, resulting in a focused output pattern. The pallidal activity changes are conveyed to the targets in thalamus (VLo) and midbrain (MEA), causing disinhibition of neurons involved in the desired motor program and inhibition of” surrounding neurons involved in competing motor programs. Abbreviations — GPi: globus pallidus, pars interna; MEA: midbrain extrapyramidal area; STN: subthalamic nucleus; VLO: ventral lateral thalamic nucleus, pars oralis. Excitatory projections are indicated with open arrows inhibitory projections are indicated with filled arrows. Relative magnitude of activity is represented by line thickness.

of MPGs involved in maintaining the body upright against gravity cannot be inhibited for the reaching arm, the voluntary movement of.that arm will meet resistance from mechanismsthat are trying to keep it in place. The reach may still be accomplished, but only slowlyand with great effort since the voluntary activation of agonists in the reach would have to overcome the postural activity of antagonists. In addition to the inhibition of mechanisms that are active prior to a voluntary movement, in many cases it is also necessary to inhibit mechanisms that might be activated by the movement itself. Consider the transcortical stretch reflexwhich acts to maintain limb position against displacement. If the arm is displaced from a maintained position, the transcortical stretch reflexis activated to resist the perturbation and return the arm to its initial position. During a normal reaching movement, the elbow is extended rapidly with the assistance of gravity and the elbow flexor muscles are stretched. The transcortical reflex to the flexors must be inhibited during the rapid extensionor it willoppose the extensionand interfere

415

with the ability to make the movement rapidly and efficiently. As a second example, consider mechanisms that are involved in maintaining the body upright in space. These mechanisms are many, including long-latency stretch reflexes, propriospinal, and vestibulospinal mechanisms. If a subject stands on a platform and that platform is suddenly moved backwards, there is a stereotyped sequence of muscle activity in the legs and back that maintains the body upright and prevents the subject from falling. If the subject now sits and makes a voluntary movement involving ankle rotation such as depressing the accelerator of a car or the pedal of a sewing machine, activation of upright postural mechanisms by the ankle rotation would interfere with the desired movement and might prevent it from occurring altogether. If the tonically active basal ganglia output inhibits competing motor mechanisms, the tonic inhibition must be removed from desired mechanisms. This must be done in a focused manner at the right time and in the right context in order not to activate competing mechanisms. The vast machinery of the striatum with its context-specificity, plasticity and spatiotemporal filtering selects which MPGs should be allowed to turn on. Thus, when a movement is made, the basal ganglia output has two parallel actions: inhibition of a multitude of competing MPGs via STN and GPi and focused selection of desired MPGs via striatum and GPi. Dysfunction of either of these actions would cause abnormal posture and movement.

5.2. The HypothesisCan Accountfor the Syndromeaof Basal Ganglia Dysfunction Many of the movement abnormalities that result from basal ganglia lesionscan be predicted from this proposed scheme of normal function. These will be discussed in relation to pallidal ablation, chorea/ hemiballismus,and parkinsonism. 5.2.1. Pallidal Ablation When the pallidum is ablated, the inhibitory GPi output is eliminated and all target mechanisms are disinhibited. Although inhibition is removed from desired motor mechanisms, it is also removed from competing mechanisms. The movement initiatory mechanisms in cerebral cortex are intact, but the competition between unwanted programs leads to slowing of movement and to excess muscle activity (cocontraction) resulting in abnormal postures. This is similarto what is seenclinicallyin torsion dystonia. While dystonia is seen after some pallidal lesions in humans, it is not known whether decreased basal gangliaoutput is an underlyingmechanismof torsion dystonia. 5.2.2. Chorea/Hemiballismus The lesion that most consistently causes chorea or

hemiballismus is in the subthalamic nucleus. When the subthalamic nucleus is ablated, there is loss of

416

J. W. Mink

both tonic and movement-related phasic excitatory input to GPi. The loss of excitatory input to GPi

results in an inability to inhibit unwanted or competing MPGs which is results in involuntary movements. This condition differs from that of pallidal ablation by the degree of reduction of GPi activity. After STN ablation, the GPi output is slightlydecreased,but after pallidal ablation, the GPi output is absent. As was discussed above, it is unlikely that tonically reduced GPi output alone is sufficientto produce chorea. More likely, the loss of excitatory input to GPi results in abnormal phasic or bursting activity in GPi or its targets and this activity causes the phasic involuntary movements of chorea. After STN ablation, the inhibitory projection from striatum to GPi remains intact. Therefore, inhibition can be removed from the desired MPGs and the intended voluntary movements themselves can be generated despite the superimposedchorea.

5.2.3. Parkinsonism Until recently,it was assumed that the GPi output was abnormally low in Parkinson’s disease (Mink and Thach, 1991c;Penneyand Young, 1983).In part, this was because of the similarity between the symptoms of Parkinson’s disease and GPi lesions, including slowness of movement, abnormal flexed posture, and cocontraction rigidity. However, as discussed above, it is now thought that GPi activity is slightly increased in Parkinson’s disease. From increasedGPi activity,one wouldpredict an excessive inhibition of all MPGs resulting in an inability to completely turn on those involved in the desired movement. The result would be slowing of movement, decreasedamplitude of movement,and in some cases an inability to move at all. From a tonically increased GPi output, one might expect that competing motor programs would be inhibited so that there would not be abnormal postures or cocontraction rigidity. However, rigidity and abnormal postures are a major component of parkinsonism, suggesting that there must be another abnormality in addition to tonically increased GPi activity. We know that in animal models of parkinsonism there is abnormal phasic activity of GPi neurons in response to somatosensory stimulation, electrical stimulation of the striatum, and at rest (section4.2.6). Unfortunately, we do not known how GPi neurons discharge in relation to voluntary movementin parkinsonism.Although there is a slight tonic increase in GPi activity, there may also be an inability to increase the GPi output appropriately during movement to inhibit competing MPGs. If so, this would explain the abnormal postures, cocontraction rigidity, and inability to suppress unwanted postural reflexesin parkinsonism.Thus, accordingto the present hypothesis,the abnormality in parkinsonism is two-fold. First, there is an inability to remove inhibition from desiredMPGs and second,there is an inability to fully inhibit competing MPGs. The proposed inability to inhibit competing MPGs in parkinsonism could be tested by recording GPi neurons during movement in monkeys with MPTP parkinsonism.

5.3. The Scheme of Focused Selectm “ n and Inhibition of Competing Motor Mechanisms and Its Relationship to other Recent Models of Basal Ganglia Fonction Over the past 30 years, many different hypotheses of basal ganglia function have been proposed (see Mink and Thach, 1991a)).Of these, a few are still popular including: (1) initiation of movement; (2) scaling of movement size and speed; (3) automatic execution of movement sequences. The scheme proposed in this review of focused selection and inhibition of competing motor mechanisms can account for many of the findings that are central to these hypothesesand also explainssomefindingsthat are inconsistent with these other hypotheses. 5.3.1. Initiation of Movement by the Basal Ganglia

An old hypothesisthat is still popular in textbooks today states that the basal ganglia initiate movement. This model was based in large part on the manifestations of human basal ganglia diseases.The paucity and slowness of movement in Parkinson’s diseasehave been attributed to an inability to initiate movementsand the involuntarymovementsof chorea and hemiballismhave been attributed to a release of otherwise normal motor systems from basal ganglia control. This model gained support from the fact that much of the output from the basal ganglia goes to parts of thalamus that project to premotor and motor cortex. Based on the anatomy and clinical phenomenology,it was argued that motor programs are stored in the basal ganglia and are called up and sent to motor cortex for execution. While some of the more dramatic deficitsof severe Parkinson’s disease (e.g.freezingand akinesia)appear to support a deficit in movementinitiation, a large body of evidencehas suggestedthat movement initiato~ mechanisms are intact in basal ganglia disease. The fact that most movement-relatedbasal ganglia neurons fire after the agonist muscles become active argues against movement initiation by the basal ganglia. Furthermore, in animals with ablation of the basal ganglia output and in many subjects with Parkinson’s disease,reaction time is normal. Nonetheless,certain basal ganglia lesions can prolong reaction time, particularly those involving the nigral dopamine neurons or striatum. If the basal ganglia do not initiate movement,how can certain basal ganglia lesions prolong reaction time? There are several possible explanations that would be consistentwith the present scheme.First, it is possible that disruption of the dopamine input to frontal cortex is responsiblefor the delayedinitiation. This could account for delayedinitiation with lesions of SNpc, but not lesionsof striatum. Second, lesions of SNpc could result in a loss of focusing by the striatum and result in an inability to inhibit foci in GPi thereby preventing release of the “brake” from desired MPGs. If motor cortex activity is recorded, one would expect to see onset of activity at the normal time, but that the magnitude of that activity would be less due to excessiveinhibition from GPi. The resulting reduced magnitude of activity might not generate enough force rapidly to overcome the

The Basal Ganglia inertial and. viscoelastic properties of the body and the mechanical onset of movement would be delayed. Thus, the mechanical onset of movement might be delayed even though the initiatory mechanisms are intact. Third, lesions of anterior putamen or caudate in the regions of neurons with set-related activity could also delay movement onset by preventing the disinhibition of movement preparatory areas in prefrontal or premotor cortex. The deficit would not lx due to the loss of motor planning mechanisms,but

to an inability to remove inhibition from those mechanisms selectively.The latter two mechanisms are consistent with the proposed scheme and are readily testable. In addition to recording the time of EMG onset after SNpc or striatal lesions, one could record the timing of activity in GPi, thalamus, or cortical targets of GPi before and after lesions. One could record from one sample of neurons before the lesion and from another sample after. A better alternative would be to record from the same single neuron in the downstream structures before and during temporary inactivation of the striatum. This would allow testing of effectof lesions on individual neurons and many neurons could be tested over time. It would be more difficultto focally inactivate SNpc without affecting SNpr, but one could inject specific dopamine antagonists into the striatum in a similar paradigm. It would be predicted that after striatal inactivation or D1 receptor blockade, the 30°/0of GPi neurons that normally decreaseactivity in relation to movement would no longer decrease. Further, it would be expected that the magnitude of activity change in premotor, supplementarymotor, or motor cortex would be reduced, but that the time of change would be the same after striatal inactivation.

417

and GPi in monkeys with chorea. While these findings are consistent with the opponent parallel pathway hypothesis, othersare not. For example, in chorea the difficultyis not that movements are too fast. Indeed, people with chorea (decreased GPi output) have slowingof voluntary movement that is similar to that seen in Parkinson’sdisease (increased GPi output). Furthermore, decreased GPi output following GPi lesions causes slowing of movement, not increased speed or size of movement. There are several similarities between the hypothesisof opponent parallel pathways and the hypothesis of focused selection and inhibition of competing motor programs. However, there is a critical difference.The opponent parallel pathway hypothesis views the role of the inhibitory and excitatory pathways to GPi to be the scaling of parameters for the intended movement in a “push-pull” manner. In contrast, the schemeproposed here viewsthe role of the inhibitory input to GPi to be the selectionof the desiredmotor programs and the role of the excitatory input to GPi to be the inhibition of competing programs with neither being involved in the scaling of specificparameters of movement. Several lines of evidence favor the scheme of focused selection and inhibition of competing motor mechanisms as has already been discussed(sections5 and 5.1).However, a specifictest of the two hypotheses has not been performed.One test of these hypotheseswould be the injection of a very small volume of muscimol into many different sites in GPi and measure the effectof

STRIATUM f

\

5.3.2. Do the Basal Ganglia Scale Movement by Opposing “Direct” and “Indirect” Pathways?

A recent popular hypothesis states that the basal ganglia circuitry is made up of opposing parallel pathways that adjust the magnitude of the inhibitory GPi output in order to increase or decrease movement (Fig. 15)(Alexanderand Crutcher, 1990a; DeLong, 1990). According to this hypothesis, increased GPi output slows movements and decreased GPi output increases movement. This hypothesis emphasizes two major paths of information flowfrom the striatum to GPi and SNpr. One is an inhibitory “direct” pathway from striatum to GPi/SNpr. The other is a net excitatory “indirect” pathway from striatum to GPe (inhibitory), GPe to STN (inhibitory),and STN to GPi/SNpr (excitatory). The two pathways are thought to be in balance such that increased activity in the “direct” pathway causes decreased GPi/SNpr output and increased activity in the “indirect” pathway causes increased GPi/SNpr output. By the adjusting the balance, the activity of cortical targets of the basal ganglia can be modulated up or down. The hypothesis predicts that decreased GPi activity would result in movements that are fast and large and that increasedGPi activitywould result in movements that are slow and small. The primary evidence for this hypothesis are the findings of increased activity in STN and GPi in monkeys with MPTP parkinsonism and decreased activity in STN

\ B“re.7t PatAway

Indirect Pathway ~

GPE

v

GPI

(

A

Fig. 15. Schematicrepresentation of the proposed “direct” and “indirect” pathways from striatum to globus pallidus pars interna. Filled symbols represent inhibitory neurons and projections;open symbolsrepresent excitatory neurons and projections.Abbreviations— GPE:globuspalliduspars externa; GP,: globus pallidus pars intema; STN: subthalarnicnucleus;MEA: midbrain extrapyramidalarea (modified from Alexander and Crutcher, 1990a)).

418

J. W. Mink

each injection on voluntary arm movement. If the opponent parallel pathway hypothesis is correct, it would be expected that (1) injection into the “arm” area would increase the amplitude or velocityof the voluntary movement and (2) injection into other areas might cause involuntary movements of other body parts but would not interferewith the voluntary arm movement. If the schemeof focused selectionis correct, it would be expected that (1) injection of muscimol into some areas would disinhibit mechanisms that compete with the arm movement causing slowingof the arm movementwith cocontraction, (2) injection into other areas might disinhibit mechanismsthat competewith postural mechanisms,causing involuntary movement from a maintained position, but (3) no injection would increase the amplitude or speed of the voluntary arm movement.A further test would be to record the activity of motor cortical neurons before and after inactivation of GPi to see if neurons that were not active during the task in normal animals becomeactiveafter GPi inactivation. This would support the focused selection and inhibition scheme.

from one behavior to another comes from the work of Cools (Cools, 1980;van den Bercken and Cools, 1982). Cook showed that blockade of dopamine transmission or injection of a cholinergicagonist in the striatum impaired the ability of animals to switch from one behavior to another. Complex movement sequences have not been studied in monkeys with focal basal ganglia lesions, but rapid sequential alternation of flexion and extension of the wrist is impaired by GPi ablation to the same degree as individual flexion and extension movements (Mink and Thach, 1991c). Brotchie et al., 1991c)and Marsden (Marsden, 1987; Marsden and Obese, 1994)have proposed a mechanism for the automatic sequencing of movement by the basal ganglia. They have suggestedthat prior to each component of a movement sequence, there is a maintained “set” or preparatory signal in SMA. Before the next component of a sequencecan be executed,the set signalmust be turned off to allow the movement to occur. They proposed that GPi neurons increase firing prior to the second and subsequent components of a movement sequenceto turn off the set signalsin SMA in order to trigger the subsequent components. This hypothesis is readily 5.3.3. Do the Basal Ganglia Sequence Movement? testable by recording in SMA before and after GPi Another popular recent hypothesisis that the basal inactivation in animals trained to perform a ganglia are responsiblefor the automatic executionof sequential movement task. learned movement sequences(Brotchie et al., 1991c; A role of the basal ganglia in sequentialmovement Marsden, 1987).This hypothesis states that mechan- is consistent with the hypothesisof focused selection isms outside the basal ganglia initiate the first and inhibition of competing motor programs component of a sequence,but that the basal ganglia proposed here. However,the mechanismis somewhat contain mechanisms for the completion of the different and is more general than that proposed by sequence. It has been shown that patients with Marsden and Brotchieet al. For voluntarymovement Parkinson’s disease have greater difficulty moving to occur, MPGs involved in the desired movement body parts simultaneously or sequentially than one must be activated and competing mechanisms must would expectfrom a.simpleaddition of the deficitsof be inhibited.For each newcomponent of a sequential each component of the movement (Benecke et al., movement, new MPGs must be turned on and the 1986; Marsden and Obese, 1994). An example of previously active MPG as well as other potentially apparent difficulty with sequential movements in competing MPGs must be inhibited. If the ability to Parkinson’s disease is the phenomenon of mi- focally select and inhibit competing motor mechancrographic (or small handwriting).This phenomenon isms is impaired, each component of a movement is common and is characterized by nearly normal- sequence would be slow and pauses between ized writing initially, but within several letters the components might be prolonged. Depending on the writing gets progressivelysmaller so that by the end movement and mechanismsinvolved, the number of of the line, it may be illegible.The early components mechanisms competing with the desired movement of the sequence are larger and faster than are the may increase additively as the sequence progresses subsequent components. As indicated above, the leading to progressiveslowing of the movement. performance of movement sequences has been studied extensivelyin peoplewith Parkinson’sdisease and some, but not all, have difficulty performing sequentialmovements.However,it is quite clear that ‘6. CONCLUSION people with Parkinson’s disease move slowly even When voluntary movement is generated, motor when they are not performingsequencedmovements. The experiment of Brotchie et al., 1991c)is often areas in the cerebral cortex send a corollary signal to cited as experimentalevidencefor a role of the basal the subthalamic nucleus which causes widespread ganglia in movement sequencing(see section 3.6.3.). excitation of GPi and SNpr and subsequent As described above, they found that most GP inhibition of motor pattern generators for competing neurons fired after the onset of movement, but that postures and movements.Simultaneously,motor and some fired before the second component of a two other areas of cerebralcortex send signalsto striatum movement sequence. The activity preceding the which filters and transforms those signals in a second component was greater if the second context-dependentmanner and then focally inhibits component was predictable, but it was always less GPi and SNpr to removetonic inhibition from motor than the activity that followed the onset of each pattern generators involvedin the desiredmovement. component of the movement. Further support for In this way, the output of the basal ganglia acts basal ganglia involvementin sequencingor switching focally to select desired motor mechanisms and

The Basal Ganglia

broadly inhibit competing motor mechanisms to allow movement to proceed without interference. Acknowledgemerrts-I

thank my mentor, W.T. Thach, for

his support and contribution to the development of these ideas.

REFERENCES

419

Bargas,J., Gallarraga, E. and Aceves,J. (1989)An early outward conductance modulates the tiring latency and frequency of neostnatalneuronsof the rat brain. .Exp. Brain Res. 75, 14&156, Baron, M., Vitek, J., Turner, R,, Kaneoke, Y., Bakay, R, and DeLong, M. (1993) Lesionsin the aensorimotorregion of the internal segment of the globus pallidus (GPi) in Parkinsonian patientsareeffectivein alleviatingthecardinalsignsof Parkinson’s disease..SOC.Neurosci. Absrr, 19, 15S4. Beaubaton,D., Amato, G., Trouche,E. and E. L. (1980)Effectsof putamencoolingon the latency,speedand accuracyof a pointing movementin the baboon. Brain Research 1%, 572-576. Benazzouz,A., Gross,C., Dupont, J. and Bioulac,B. (1992)MPTP induced hemiparkinsonismin monkeys: behavioral, mechanographic, electromyographicand immunohistochemicalstudies.

Adams, R. D. and Victor, M. (1993) Principles of Neuro/ogy, McGraw-Hill:New York. Afsharpour, S, (1985) Topographicalprojections of the cerebral cortex to the subthalamicnucleus.J. Comp, Neurol. 236, 1428. Exp. Brain Res. 90, ll&120. Agostino, R,, Berardelli, A., Formica, A., Accornero, N. and Benecke,R., Rothwell,J. C., Dick,J. P. R., Day,B. L. and Marsden, Manfredi,M. (1992)Sequentialarm movementsin patientswith C, I). (1986)Performanceof simultaneousmovementsin patients Parkinson’sdiseaseHuntington’sdiseaseand dystonia.Brain 11S, with Parkinson’sdisease, Brain 109,73>757. 1481-1495. Benecke,R,, Rothwell,J. C., Dick,J, P. R., Day, B.L. and Marsden, Agosdno, R.,Berardelli, A.,Formica, A.,Stocchi, F.,Accornero, N. C. D, (1987)Simpleand complexmovementsoffand on treatment andManfredi, M,(1994) Analysis ofrepetitive andnonrepetitive in patients with Parkinson’s disease. J. Neuro/, Neurosurg.

sequential armmovements in patientswithParkinson’s disease. Mov. Disord. 9, 311-314.

Akins, P. T., Surmeier,D. J. and Kitai, S. T. (1990)Muscarinic modulation of a transient K + conductancein rat neostriatal neurons. Nature 344, 240-242. Albin, R. L., Young,A. B, and Penney,J. B. (1989)The functional anatomyof basalgangliadisorders. Trends AJeurosci,12,366-375. Alexander,G. and Crutcher, M. (1991)Parallel processingin the basal ganglia:up to a point. Trends Neurosci.14, 56-5S. Alexander,G. E. (1987) Selectiveneuronal dischargein monkey putamen reflectsintendeddirectionof planned limb movements. Exp, Brain Res. 67, 623634.

Alexander, G. E. and Crutcher, M. D. (1990a) Functional architectureof basal gangliacircuits:neuralsubstratesof parallel processing.Trends Neurosci,13, 26&271. Alexander, G. E. and Crutcher, M. D. (1990b)Preparation for movement:Neural representationsof intendeddirectionin three motor areas of the monkey.J. Neurophysio/. 64, 133–150. Alexander,G. E. and DeLong,M. R. (1985)Microstimulationof the primate striatum. 11. Somatotopic organization of striatal microexcitablezones and their relation to neuronal response properties. J, Neurophysio/, 53, 1417–1430. Alexander,G. E., DeLong,M. R. and Strick, P, L. (1986)Parallel organization of functionally segregated circuits linking basal gangliaand cortex. Arm, Reu. Neumsci. 9, 357–381. Amato, G., Trouche, E., Beaubaton,D. and Grangetto, A. (1978) The role of internal pallidal segmenton the initiation of a goal directedmovement.Neurosci.Left. 9, 159-163. Anderson,M. E. and Horak, F. B. (1985)Influenceof the globus pallidus on arm movements in monkeys 111. Timing of movement-relatedinformation.J. Neurophysiof. 54, 433448. Anderson, M. E., Inase, M., Buford, J. and Turner, R. S. (1993) Movementand preparatoryactivityof neuronsin paOidal-receiving areas of the monkeythalamus.In: Role of fhe cerebe//um and basal ganglia in voluntarymovement.Eds. N. Mano, 1. Hamada and M. R. DeLong. ElsevierSciencePublishers:Amsterdam. Aosaki,T., Kimura, M. and Graybiel,A. M. (1995)Temporaland spatial characteristicsof tonicallyactiveneuronsof the primate’s striatum. J. Neurophysiol. 73, 123+1252. Aosaki,T., Tsubokawa,H,, Ishida,A., Watanabe,K., Graybiel,A. M. and Kimura, M. (1994)Responsesof tonicallyactiveneurons in the primate’s striatum undergo systematicchanges during behavioralsensorimotorconditioning.J. Neurosci.14,396%3984. Apicella,P., Ljungberg,T., Scarnati, E. and Schultz, W. (1991a) Responsesto rewardin monkeydorsaland ventralstriatum. Exp. Brain Res. 85, 491-500.

Apicella, P., Scarnati, E., Ljungberg,T. and Schultz, W. (1992) Neuronal activityin monkeystriatum related to the expectation of predictableenvironmentalevents.J. Neurophysiol. 68,945-960. Apicella,P., Scamati, E. and Schultz, W. (1991b) Tonically dischargingneurons of monkeystriatum respond to preparatory and rewardingstimuli. Exp, Brain Res. 84, 672-675. Augood, S. J., Emson, P. C., Mitchell,1.J., Boyce,S., Clarke, C. E. and Crossman,A. R. (1989)Cellularlocalizationof Enk. gene expressionin MPTP-treated cynomolgusmonkeys. Mol. Brain Res. 6, 85-92.

Psychiatr. 50, 296-303.

Berardelli,A., Sabra, A. F. and Hallett, M. (1983)Physiological mechanisms of rigidity in Parkinson’s disease. J. Nerwol, Neurosurg.Psychiatr. 46, 4>53. Bergman,H., Wicbmann,T. and DeLong,M. R. (1990)Reversalof experimentalparkinsonismby lesionsof the subthalamicnucleus. Science 249, 1436-1438.

Bergman,H., Wichmann,T., Karmon,B,and DeLong,M. R, (1994) The primate subthalamicnucleus. II. Neuronal activity in the MPTP madel of Parkinsonism.J. Neurophysiol. 72, 507-520. Bhatia,K. P. and Macsden,C. D. (1994)The behavioral and motor consequencesof focal lesionsof the basal gangliain man. Brain

117,859876. Bolam,J. P. andSmith,Y. (1992) Thestriatumand the globus

pallidussend convergentsynapticinputs onto singlecellsin the entopeduncularnucleusof the rat: A doubleanterogradelabeling study combinedwith postembeddingimmunocytochemistryfor GABA. J. Comp. Neurol. 321,456476. Bouyer,J. J., Park, D. H., Joh, T. H. and Pickel, V. M. (1984) Chemicaland structural analysisof the relation betweencortical inputs and tyrosine hydroxylase-containingterminals in rat neostriatum. Brain Res. 3t32, 267–275. Bradshaw, J. L., Phillips, J. G., Dennis, C., Mattingley, J. B., Andrewes,D., Chiu,E., Pierson,J. M. and Bradshaw,J. A. (1992) Initiationand exeeutionof movementsequencesin thosesuffering fromand at-riskof developingHuntington’sdisease.J. Clin. Exp. Neuropsych. 14, 179-192.

Brooks,D. J. (1995)The role of the basalgangliain motor control: contributionsfrom PET. J. Neurol. Sci. 128,1-13. Brotchie,J. M., Mitchell,1.J., Sambrook,M. A. and Crossman,A. R. (1991a) Alleviation of parkinsonism by antagonism of excitatoryamino acid transmissionin the medialsegmentof the globus pallidusin rat and primate. Moo. Disord. 6, 133–138. Brotchie,P., Iansek,R. and Home, M.K. (1991b)Motor function of the monkey globus pallidus. 1. Neuronal discharge and parametersof movement.Brain 114, 1667-1683. Brotchie,P., Iansek,R. and Home, M. K. (1991c)Motor function of the monkeyglobuspallidus.2. Cognitiveaspectsof movement and phasicneuronal activity. Brain 114, 1685-1702. Brown, R. G., Jahanshahi, M. and Marsden, C. D. (1993)The executionnf bimanual movementsin patients with Parkinson’s, Huntington’sand cerehellardisease [seecomments].Journa/ of Neurology, Neurosurgery and Psychiatry 56, 295-297.

Bruce,C. J. and Goldberg,M. E. (1985)Primatefrontal eye fields. I Singleneuronsdischargingbeforesaccades.J. Neurophysio/. 53, 603-635.

Caan, W., Perrett, D. L and Rolls,E. T. (1984)Responsesof striatal neurons in the behaving monkey. 2. Visual processingin the caudal neostriatum. Brain Res. 290, 53-65. Camarata, P. J., Parker, R. G., Park, S.-K., Haines,S. J., Turner, D. A., Chae, H. and Ebner, T. J. (1992) Effects of l-methyl-4-phenyl-l,2,5,6-tetrahydropyridine(MPTP)-induced hemiparkinsonismon the kinematics of a two-dimensional, multijointarm movementin the rhesusmonkey.Neuroscience 48, 607-619.

Campbell, G. A., Eckhardt, M. J. and Weight, F. (1985)

420

J. W. Mink

Dopaminergicmechanismsin suhthalamicnucleusof rat: analysis using horseradishperoxidaseand microiontophoresis.Brain Res. 333, 261-270.

Canavan,A. G. M., Passingham,R. E., Marsden,C. D., Quinn,N., Wyke, M. and Polkey, C. E, (1989) Sequencing ability in Parkinsonians,patientswith frontal lobelesionsand patientswho have undergoneunilateraltemporalIobectomies.Neuropsychologia 27, 787-798.

Carpenter,M, B. (1981)Anatomyof the corpus striatum and brain stem integrating systems. In: Handbook of Physiology: The Neruous Sys@m Vol. 2, pp. 947–995. Ed. V. B. Brooks.American PhysiologicalSociety:Bethesda. Carpenter, M. B. and Carpenter, C. S. (1951) Analysis of somatotopicrelationsof the corpus Luysiin man and monkey.J. Comp. Neurol. 95, 349-370. Carpenter, M. B., Nakano, K. and Kim, R. (1976)Nigrothalamic projections in the monkey demonstrated by autoradiographic techniques.J. Comp. Neurol. 165,401416. Carpenter, M. B, and Whittier, J. B. (1952)Study of methods for producingexperimentallesionsof the centralnervoussystemwith specialreferenceto the stereotaxictechnique.J. Corrrp.Neuro/. 97, 73-131. Carpenter,M. B., Whittier,J, R. and Mettler,F. A, (1950)Analysis of choreoid hyperkinesiain the rhesus monkey. Surgical and pharmacologicalanalysisof hyperkinesiaresultingfrom lesionsin the subthalamicnucleusof Luys.J. Comp. Neurol. 92, 293-331. Cherubini,E., Herrling,P. L., Lanfrrmey,L. and Stanzione,P. (1988) Excitatory amino acids in synaptic excitation of rat striatal neurones in oitro. J. Physial. 400, 677-690. Conde, H., Benita,M., Dormont, J. F., Schmied,A. and Cadoret, A. (1981) Control of reaction time performance involvesthe striatum. J. Physio/. (Paris) 77, 97–105, Cools, A. R. (1980)Role of the neostriataldopaminergicactivityin sequencingand selectingbehavioral strategies:Facilitation of processesinvolved in selectingthe best strategy in a stressful situation. Behav. Brain Res. 1, 361–378. Cooper, 1. S. and Bravo, G. (1958) Chemopallidectomyand chemothalamectomy.J. Neurosurg.15, 244-250. Corcos, D. M., Chen, C.-M., Quin, N. P., McAuley, J, and Rothwell,J. C. (1996)Strengthin Parkinson’sdisease:relationship to rate of force generationand clinicalstatus. Ann. Neurol. 39, 79-88.

Cowan, R. L. and Wilson,C. J. (1994)Spontaneousfiringpatterns and axonal projectionsof singlecorticostriatalneuronsin the rat medialagranrdarcortex. J. Neurophysio/. 71, 17–32. Crammond, D. J. and Kalaska, J. F. (1994) Modulation of preparatory neuronal activity in dorsal premotor cortex due to stimulus-responsecompatibility.J. Neurophysiol. 71, 1281-1284. Crossman, A. R. (1987) Primate models of dyskinesia: The experimental approach to the study of basal ganglia-related involuntarymovementdisorders. Neuroscience 21, I-40. Crossman,A. R., Clarke, C. E., Boyce,S., Robertson, R. G. and Sambrook, M. A. (1987) MPTP-inducedParkinsonismin the monkey: Neurochemicalpathology,complicationsof treatment and pathophysiologicalmechanisms.Can. J. Neurol. Sci. 14, 428+35, Crossman, A. R., Sambrook, M. A. and Jackson, A. (1984) Experimentalhemichorea/hemiballismus in the monkey.Studies on the intracerebralsite of action in a drug-induceddyskinesia.

the monkeyproduced by kainic acid: histologicand behavioral effects.Appl. Neurophysiol. 42, 95=97. DeLong,M. R. and Strick, P. L. (1974)Relationof basal ganglia, cerebellum,and motor cortex units to ramp and ballistic limb movements,Brain Res. 71, 327–335. DeLong,M. R., Crutcher, M. D. and Georgopordos,A. P. (1983) Relations between movement and single cell discharge in the substantial nigra of the behaving monkey. J. Neurosci. 3, 159%1606,

DeLong, M, R,, Crutcher, M, D. and Georgopoulos,A. P. (1985) Primate globus pallidus and subthalamic nucleus: functional organization.J. Neurophysiol. 53, 53W543, DeLong,M, R, and Georgopordos,A, P. (1981)Motor functionsof the basal ganglia. In: Handbook of Physiology, The Nervous System Vol. 2, pp. 1017–1062.Ed. V. B. Brooks. American PhysiologicalSociety:Bethesda, Denny-Brown,D. (1962) The Basal Ganglia and Their Relation to Disorders of Mouemenr. Oxford UniversityPress: London. Denny-Brown,D. (1967)The fundamentalorganizationof motor behavior. In: Neurophysiological Basis of Normal and Abnormal Mo[or Acrioities, pp. 415442. Eds. M. Yahr and D. Purpura. Raven Press:New York, Denny-Brown,D. and Yanagisawa,N. (1976)The role of the basal ganglia in the initiation of movement.In: The Basal Gang/ia, pp. 11$149. Ed. M. D. Yahr. Raven Press:New York. Deuschl,G., Lucking,C. H. and Schenck,E. (1989)Hand muscle reflexesfollowingelectrical stimulation in choreatic movement disorders..?, Neurol. Neurosurg. Psychiatr. 52, 75>762. DeVito, J. L. and Anderson, M. E. (1982)An autoradiographic study of efferentconnectionsof the globus pallidus in Macaca mulatta. Exp. Brain Res. 46, 107–117, Di Chiara, G., MorelIi, M, and Consolo, S. (1994) Modulatory functionsof neurotransmittersin the striatum: ACh/dopamine/ NMDA interactions, Trends Neurasci, 17, 228-233. DiFiglia, M. and Rafols,J. A. (1988)Synapticorganizationof the globus pallidus. J. J3k?ctr.A4icrosc. Tech. 10, 247-263. Doudet, D. J,, Gross, C., Arluison, M. and Bioulac, B. (1990) Modificationsof precentralcortexdischargeand EMG activityin monkeyswith MPTP-inducedlesionsof DA nigralneurons. Exp. Brain Res. 80, 177-188.

Evarts,E, V. and Tanji,J. (1974)Gatingof motor cortexreflexesby prior instruction. Brain Res. 71, 479-494. Evarts, E. V. and Tanji, J. (1976)Reflexand intendedresponsesin motorcortexpyramidaltract neuronsof monkey.J, Neurophysial. 39, 10691080.

Evarts,E. V., Teravainen,H. and Cahre,D. B. (1981)Reactiontime in Parkinson’sdisease. Brain 104,167–186. Feger, J. and Crossman, A. R, (1984) Identificationof different subpopulations of neostriatal neurones projecting to globus pallidus or substantial nigra in the monkey: A retrograde fluorescencedouble-labelingstudy. Neurasci. Letr. 49, 7-12. Filion, M. and Tremblay,L. (1991)Abnormalspontaneousactivity of globus pallidus neurons in monkeys with MPTP-induced parkinsonism.Brain Res. 547, 142-151. Flaherty, A. W, and Graybiel, A. M. (1991) Corticostriatal transformationsin the primatesomatosensorysystem.Projections from physiologicallymapped body-part representations. J. Neurophysiol. 66, 1249-1263.

Flaherty,A. W. and Graybiel,A. M. (1993a)Output architectureof the primate putamen. J. Neurosci. 13, 3222–3237. Brain 107, 579–596. Crutcher, M. D. and Alexander,G. E. (1990)Movement-related Flaherty,A. W. and Graybiel,A. M. (1993b)Two input systemsfor body representationsin the primate striatal matrix:experimental neuronal activity selectivelycoding either direction or muscle evidencein the squirrelmonkey. J. Neurosci. 13, 1120-1137. pattern in three motor areas of the monkey.J. Neurophysio/. 64, Flaherty, A, W. and Graybiel, A. M. (1994) Input-output 151-163. organizationof the sensorimotorstriatumin the squirrelmonkey. Crutcher, M. D. and DeLong,M. R. (1984a)Singlecell studiesof J. Neurosci. 14, 599-610. the primate putamen 1.Functionalorganization.Exp. Brain Res. Flash, T., Inzelbcrg,R., Schechtman,E. and Korczyn,A. D. (1992) 53, 233-243. Kinematic analysis of upper limb trajectories in Parkinson’s Crutcher, M. D. and DeLong,M. R. (1984b)Singlecell studiesof disease. Exp. Neurol. 118,21S226. the primateputamen.11.Relationsto directionof movementsand Flowers, K. A. (1976) Visual “closed-loop” and “open-loop” patterns of muscularactivity. Exp. Brain Res. 53, 244258. characteristicsof voluntarymovementin patientswith ParkinsonDeLong, M. R. (1971) Activity of pallidal neurons during ism and intention tremor. Brain 99, 269-310. movement.J. Neurophysiol. 34, 41U27. DeLong,M. R. (1973)Putamen:Activityof singleunitsduringslow Francois, C., Percheron,G., Yelnik, J. and Heyner, S. (1984)A Golgi analysis of the primate globus pallidus. I. Inconstant and rapid arm movements.Scierrce 179, 1240-1242. processesof large neurons, other neuronal types, and afferent DeLong, M. R. (1990)Primate modelsof movementdisordersof axons. J. Comp. Neurol. 227, 182-199, basal gangliaorigin. TrerrdrNeurosci. 13, 281-285. DeLong, M. R. and Coyle,J. T. (1979)Globus palliduslesionsin Francois, C., Percheron,G., Yehrik,J. and Tande, D. (1988)A

The Basal Ganglia topographic study of the course of nigral axons and of the distribution of pallidal axonal endings in the centre median-parafaacicularcomplex of macaques. Brain Res. 473, 181-186.

Freeman,J. S., Cody, F. W. J. and Schady,W. (1993)The influence of externaltimingcues upon the rhythmof voluntarymovements in Parkinson’s disease. J. Neurol. Newcmurg. Psychiatr. 56, 1078-1084.

Fujimoto, K. and Kita, H. (1993) Response characteristics of subthalamicneuronsto the stimulationof the aensorimotorcortex in the rat. Brain Res. 609, 185-192. Gaspar, P., Duyckaerts,C., Alvarez,C., Javoy-Agid,F. and Berger, B. (1991) Alterations of dopaminergic and noradrenergic innervationsin motor cortexin Parkinson’sdisease.Ann. Neurol. 30, 365-374.

Gaspar, P., Stepniewski,I. and Kaas, J. H. (1992)Topographyand collateralizationof the dopaminergicprojectionsto motor and lateral prefrontal cortex in owl monkeys.J. Comp. Neurol. 325, 1-21.

421

sionalreconstructionof axons,synapses,and postsynaptictargets in rat neostriatum, Neuroscience 58, 593+04. Hallett, M. and Khoshbin,S. (1980)A physiologicalmechanismof bradykinesia,Brain 103, 301-314. Hamada, 1.and DeLong,M. R. (1992a)Excitotoxicacid lesionsof the primate subthalamic nucleus result in reduced pallidal neuronal activity during active holding. J. Neurophysia[. 68, 185%1866.

Hamada, I. and DeLong,M. R, (1992b)Excitotoxicacid lesionsof the primatesubthalamicnucleusresultin transientdyskinesiasof the contralaterallimbs.J. Neurophysia/. 68, 1850-1858. Hamada, 1., DeLong, M. R. and Mano, N.-1. (1990)Activity of identifiedwrist-relatedpallidal neurons during step and ramp wrist movementsin the monkey,J. ~europhysiol. 64, 1892-1906. Hartmann-von Monakow, K., Akert, K. and Kunzle, H. (1978) Projectionsof the precentralmotor cortexand othercorticalareas of the frontallobeto the subthalamicnucleusin the monkey.E.rp. Brain Res, 33, 39W03.

Hauber,W., Bubser,M. and Schmidt,W. J. (1994)6-Hydroxydopamine lesionof the rat prefrontalcortex impairsmotor initiation but not motor execution.Exp. Brain Res. 99, 524528, Hayashi, A., Kagamihara, Y., Nakajima, Y., Narabayashi, H., Okuma, Y. and Tanaka, R. (1988) Disorder in reciprocal innervationupon initiation of voluntary movementin patients with Parkinson’sdisease.Exp. Brain Res. 70, 437440. Hazrati, L.-N. and Parent, A. (1992a) Differential patterns of arborizationof striatal and subthalamicfibersin the two pallidal segmentsin primates. Brain Res. 598, 31I–315. Hazrati, L.-N. and Parent, A. (1992b)The striatopallidalprojection displaysa high degree of anatomicalspecificityin the primate.

Georgopoulos, A. P. (1994) New concepts in generation of movement.Neurons13, 257–268. Georgopoulos,A. P., DeLong,M. R. and Crutcher, M. D. (1983) Relation between parameters of step-trackingmovementsand singlecelldischargein theglobuspallidusand subthalamicnucleus of the behavingmonkey.J. Neurosci.3, 1586-1598. Gerfen,C. R. (1988)Synapticorganizationof the striatum.J. .E/ectr. Microsc. Tech. 10, 26>281. Gerfen, C. R. (1992) The neostriatal mosaic: Multiple levels of compartmental organization in the basal ganglia. Ann. Reu. Neurosci.15, 285-320. Brain Res. 592, 213-227. Gerfen,C. R., Engber,T. M., Mahan, L. C., Susel,Z., Chase,T. N., Monsma, F. J. and Sibley,D. R. (1990)D, and Dj dopamine Hebb,D. O. (1949)The Organization of Behavior. Wiley:NewYork. receptor-regulatedgeneexpressionof striatonigraland striatopal- Hedreen, J. C. and DeLong, M. R. (1991) Organization of Iidal neurons. Science 250, 142%1432. striatopallidal,striatonigral,and nigrostriatal projectionsin the Gerfen, C. R. and Young, W. S. 111.(1988) Distribution of macaque.J. Comp. Neurol. 304, 569-595. striitonigral and striatopallidalpeptidergicneuronsin both patch Hikosaka, O., Matsumura, M., Kojima, J. and Gardiner, T. W. (1993)Role of the basal gangliain initiationand suppressionof and matrixcompartments:an in situ hybridizationhistochemistry and fluorescentretrogradetracingstudy. BrairrRes. 460, 161–167. saccadiceye movements.In: Role of the cerebellum and basal ganglia in voluntary movement, pp. 213-219.Eds. N. Mano, 1. Gimenez-Amaya,J. M. and Graybiel,A. M. (1990)Compartmental Hamada and M. R. DeLong. Elsevier Science Publishers: origins of the striatopallidal projection in the primate. NeuroAmsterdam. science34, 111–126. Gimenez-Amaya,J.-M. and Graybiel, A. M. (1991) Modular Hikosaka, 0., Sakamoto, M. and Usui, S. (1989a) Functional organizationof projectionsneuronsin the matrixcompartmentof properties of monkey caudate neurons. 1. Activitiesrelated to the primate striatum. J. Neurosci. 11, 779-791. saccadiceye movements.J. Neurophysio/, 61, 78t&798. Goetz, C. G., DeLong, M. R., Penn, R. D. and Bakay, R. A. E. Hikosaka, O., Sakamoto, M. and Usui, S. (1989b) Functional properties of monkey caudate neurons II. Visual and auditory (1993)Neurosurgicalhorizonsin Parkinson’sdisease.Neurology properties.J. Neurophysiol. 61, 799-813. 43, 1-7. Goldman,P. S. and Nauta, W. J. H. (1977)An intricatelypatterned Hikosaka, O., Sakamoto, M. and Usui, S. (1989c) Functionalpropertiesof monkeycaudate neurons.111.Activities prefronto-caudateprojection in the rhesus monkey. J. Crmp. related to expectationof target and reward.J. Neurophysio/. 61, Neurol. 171, 36%386. 814-831. Graybiel,A. M. (1990)Neurotransmittersand neuromodtdatorsin Hikosaka, O. and Wurtz, R. H. (1983a)Visual and oculomotor the basal ganglia. TrerrdrNeurosci. 13, 244-254. functionsof monkeysubstantialnigra pars reticulate.1. Relation Graybiel,A. M., Aosaki,T., Flaherty,A. W. and Kimura,M. (1994) The basal ganglia and adaptive motor control. Science 265, of visualand auditoryresponsesto saccades.J. Neurophysio/. 49, 123W1253, 182tL1831. Graybiel,A. M. and Ragsdale,C. W. (1978)Histochemicallydistinct Hikosaka, 0, and Wurtz, R. H. (1983b)Visual and oculomotor functionsof monkey substantialnigra pars reticulate. 11.Visual compartments in the striatum of human, monkey and cat responses related to fixation of gaze. J. Neurophysio/. 49, demonstratedby acetylcholinesterasestaining. Proc. Nat/. Acad. Sci. 75, 5723-5726.

Graybiel,A. M., Ragsdale,C. W., Yoneika,E. S. and Elde, R. P. (1981)An immunohistochemicalstudy of enkephalinsand other neuropeptidesin the striatum of the cat with evidencethat the opiate peptidesare arranged to form mosaicpatterns in register with striosomal compartments visible with acetylcholinesterase staining. Neuroscience 6, 377-397. Gross, C., Feger,J., Seal,J., Haramburu,P. and Bioulac,B. (1983) Neuronalactivityin area 4 and movementparametersrecordedin trained monkeysafter unilateral lesion of the substantialnigra. Exp. Brain Res. Suppl.7, 181–193. Groves, P. M., Garcia-Munoz,M., Linder, J. C., Manley, M. S., Martine, M. E. and Young, S. J. (1995)Elementsof the intrinsic organizationand informationprocessingin tbe neostriatum.In: MoaWsof Information Processing in the Basal Ganglia, pp. 51-96. Eds. J. C. Houk, J. L. Davis and D. G. Beiser. MIT Press: Cambridge. Groves, P. M., Linder, J. C. and Young, S. J. (1994) 5-Hydroxydopamine-labeled dopaminergicaxons: three dimen-

12541267.

Hikosaka, O. and Wurtz, R. H. (1983c)Visual and oculomotor functions of monkey substantial nigra pars reticulate. 111 Memory-contingentvisualand saccaderesponses.J. Neuroph},siol. 49, 1268–1284.

Hikosaka, O. and Wurtz, R. H. (1983d)Visual and oculomotor functionsof monkeysubstarrtianigrapars reticulate.IV. Relation of substantialnigra to superior colliculus.J, Neurophysiol. 49, 1285-1301.

Hikosaka,O. and Wurtz, R. H. (1985a)Modificationof saccadiceye movementsby GABA-relatedsubstances.I. EfTectof muscimol and bicucullinein monkeysuperiorcolliculus.J. Neurophysiol. 53, 266-291.

Hikosaka,0. and Wurtz,R. H. (1985b)Modificationofsaccadiceye movementsby GABA-relatedsubstances.11.Effectsof muscimol in monkeysubstantialnigra pars retictdata. J. Neurophysio/. 53, 292-308.

Homberg,V. and Huttunen,J. (1994)Muscletone in Huntington’s disease.J. Neurol. Sci. 121,147-154.

422

J. W. Mink

Hoover,J. E. and Strick, P. L. (1993)Multipleoutput channelsin the basal gmrglia.Science 259, 81%321. Horak, F. B. and Anderson, M. E. (1984) influence of globus pallidus on arm movements in monkeys. I Effects of kainic acid-inducedlesions.J. Neurophysiol. 52, 29W304. Horak, F. B., Nutt, J. G. and Nashner, L. M. (1992) Posturalinflexibilityin Parkinsoniansubjects.J. Neuro/.Sci. 111, 4658, Here, J., Meyer-Lohmaon,J. and Brooks,V. B.(1977)Basalganglia cooling disables learned arm movements of monkeys in the absenceof visualguidance..Scierrce195, 58+586. Here, J. and Vilis,T. (1980)Arm movementperformanceduring reversiblebasalganglialesionsin the monkey.Exp. Brain Res. 39, 217-228.

lnase, M. and Tanji, J. (1995)Thalamicdistributionof projection neuronsto the primarymotor cortex relativeto afferentterminal fieldsfrom the glohuspallidusin the macaquemonkey.J. Corrrp. Neurol. 353, 415-426.

Izzo, P. N. and Bolam,J. P. (1988)Cholinergicsynapticinput to differentparts of spinystriatonigralneuronsin the rat. J. Comp. Neurol, 269, 21%234. Jackson,A. and Crossman,A. R. (1984)Experimentalchoreoathetosis produced by injection of a gamma-aminobutyricacid antagonist into the Ientiformnucleus in the monkey. Neurosci. Leff. 46, 4145.

Jaeger, D., Gilman, S. and Aldridge,J. W. (1993)Primate basal ganglia activity in a precued reaching task: Preparation for movement.Exp. Brain Res. 95, 51-64. Jaeger, D., Kita, H. and Wilson,C. J. (1994)Surround inhibition among projection neurons is weak or nonexistent in the rat neostriatum. J. Neurophysiol. 72, 2555–2558. Jahanshahi, M., Brown, R. G. and Marsden, C. D. (1993) A comparative study of simple and choice reaction time in Parkinson’s, Huntington’s and cerebella disease. Journal of Neurology, Neurosurgery and Psychiatry 56, 1169-1177. Jenner, P., Rupniak, N. M. J., Rose, S., Kelly, E., Kilpatrick,G., Lees,A. and Marsden,C. D. (1984)l-methyl-4-phenyl-1,2,3,6-tetrahydropyridine-inducedParkinsonism in the common marmoset. Neurosci. Leff. 50, 8>90. Jiang, Z.-G. and North, R. A. (1991)Membraneproperties and synapticresponsesof rat striatalneuronesirruilro. J. Physiol. 443,

basal gangliain monkeysand chimpanzees.J, Neurophysiol. 7, 128-148.

Kievit, J. and Kuypers, H. G, J. M, (1977)Organizationof the tbalamo-corticalconnexionsto the frontal lobe in the rhesus monkey. Exp. Brain Res. 29, 299–322. Kimura, M. (1986)The role of primate putamen neurons in the associationof sensorystimuliwith movement.Neurosci. Res. 3, 436-443.

Kimura, M., Kate, M. and Shimazaki, H. (1990) Physiological propertiesof projectionneurons in the monkey striatum to the globus pallidus. Exp. Brain Res. 82, 672-676. Kimura, M., Rajkowski, J. and Evarts, E. (1984) Tonically dischargingputamen neurons exhibit set-dependentresponses. Proc. Natl. Acad. Sci. 81, 4998–5001.

Kita, H. and Kitai, S. T. (1988) Glutamate decarboxylaw immunoreactiveneuronsirt rat neostriatum:their morphological typesand populations. Brain Res. 447, 346352. Kitai, S. T., Kocsis,J. D., Preston, R. J. and Sugimori,M. (1976) Monosynapticinputs to caudate neuronsidentifiedby intracellular injectionof horseradishperoxidase,Brain Res. 109,60 L606. Klockgether,T. and Dichgans, J. (1994) Visual control of arm movementin Parkinson’sdisease.h40u.Disord. 9, 48–56. Kornhuber,H, H. (1971)Motor functionsof cerebellumand basal ganglia: the cerebellocortical saccadic (ballistic) clock, the cerebellonuclearhold regulator, and the basal ganglia ramp (voluntary speed smooth movement) generator. Kyberrretik8, 157–162. Kunesch,E., s-chnitzler,A., Tyercha,C., Knecht,S. and Stelmach, G, (1995)Altered force releasecontrol in Parkinson’sdisease. Behau. Brain Res. 67, 4349.

Kunzle,H. (1975)Bilateralprojectionsfromprecentralmotorcortex to putamen and other parts of the basal ganglia. An autoradiographicstudy in h4acacafmcicukzris. Brain Res. 88, 195-209.

Laitinen,L. V., Bergenheim,A. T. and Hariz, M. 1.(1992)Leksell’s posteroventral pallidotomy in the treatment of Parkinson’s disease.J. Neurosurg.76, 53-61. Langston,J. W., Ballard,P., Tetrud, J. and Irwin,1.(1983)Chronic Parkinsonismin humans due to a product of meperidine-analog synthesis.Science219, 979–980. Lapper,S. R. and Bolam,J. P. (1992)Input from the frontal cortex 533-553. and the parafascicrdarnucleusto cholinergicinterneuronsin the Johnson,T. N. and Rosvold,H. E. (1971)Topographicprojections dorsal striatum of the rat. Neuroscience51, 533–545. on the globus pallidus and the substantialnigra of selectively Lavoie,B. and Parent, A. (1994)Pedunculopontinenucleusin the placedlesionsin the precommisuralcaudatenucleusand putamen squirrelmonkey:distributionof cholinergicand monoaminergic in the monkey. Exp. Neum/. 33. neurons in the mesopontinetegmentumwith evidencefor the Kaji, R., Rothwell,J. C., Katayama, M., Ikeda, T,, Kubori, T., presenceof glutamate in cholinergicneurons. J. Corrrp. Neurol. 344, 190-209. Kohara, N., Mezaki, T., Shibasaki, H. and Kimura, J. (1995) Tonicvibrationreflexand muscleafferentblockin writer’scramp. Lidsky,T., Manetto,C. and Schnieder,J. (1985)A considerationof sensoryfactorsinvolvedin motor functionsof the basal ganglia. Ann. Neurol. 38, 155-162. Kalsner,S. and Westfall,T. C. (1990)Presynapticreceptorsand the Brain Res. Rev. 9, 13>146. questionof autoregulationof neurotransmitterrelease.Ann.N.Y. Lighthall,J. W. and Kitai, S. T. (1983)A shortdurationGABAergic inhibitionin identifiedneostriatalmediumspinyneurons: in uirro Acad. Sci. 604, 652-655. Kaerazawa,1., Kimura, M., Murata, M., Tanaka, Y, and Cho, F. slice study. Brain Res. Brdl. 11, 103–I10. (1990)Choreic movementsin the macaquemonkey inducedby Ljungberg,T., Apicella,P. and Schultz,W. (1992)Responsesof monkey dopamine neurons during learning of behavioral kainicacid lesionsof the striatumcombinedwith L-DOPA.Brain ceactions.J. Neurophysiol. 67, 145–163. 113,509–535. Kate, M. and Kimura, M. (1992)Effectsof reversibleblockadeof Marsden, C. D. (1987)What do the basal ganglia tell premotor cortical areas?. Ciba Found. Symp. 132,282–300. basalgangliaon a voluntaryarm movement.J. Neurophysiol. 68, Marsden,C. D. and Fahn, S. (1994)Movementdisorders3, pp. 529. 1516-1534. Kawaguchi,Y. (1992)Large aspiny cells in the matrix of the rat Oxford UniversityPress: Boston. neostriatum in uitro: physiologicalidentification,relation to the Marsden,C. D. and Obese,J. A. (1994)The functionsof the basal ganglia and the paradox of stereotaxicsurgery in Parkinson’s compartmentsand excitatorypostsynapticcurrents.J. Neurophysdisease. Brain 117, 877-897. iol. 67, 1669-1682. Kawaguchi,Y., Wilson,C. J. and Emson,P. C. (1989)Intracellular Marsden, C. D., Obese, J. A. and Rothwell,J. C. (1983)Clinical neurophysiologyof musclejerks: myoclonus,chorea, and tics. recordingof identifiedneostriatalpatch and matrix spinycellsin Ado. Neurrd. 39, 865-881. a slicepreparationpreservingcorticalinputs. J. Neurophysio/. 62, Marsden,C. D., Obese,J. A., Zarranz,J. J. and Lang,A. E. (1985) 1052–1068. The anatomicalbasis of symptomatichemidystonia.Brain 108, Kawaguchi,Y., Wilson,C. J. and Emson,P. C. (1990)Projection 463-483, subtypesof rat neostriatal matrix cells revealedby intracellular Marsden, C. D. and Rothwell, J. C. (1987) The physiologyof injectionof biocytin. J. Neurosci. 10, 3421-3438. idiopathicdystonia. Can. J. Neuro/. Sci. 14,521–527. Kemp,J. M. and Powell,P. S. (1971)The structureof the caudate Martin, J. P. (1967)The Basal Ganglia and Posture. J.B. Lippincott nucleusof the cat: Lightand electronmicroscopy.Phil. Trans. R. Company:Philadelphia. Soc. Lend. B 262, 383-401. Kemp,J. M. and Powell,T. P. S. (1970)Thecorticostriateprojection Matsumura, M., Kojima, J., Gardiner, T. W. and Hikosaka, O. (}992) Visual and oculomotor functions of the monkey in the monkey. Brain 93, 525-546. Kennard,M. A. (1944)Experimentalanalysisof the functionsof the subthaktmicnucleus.J. Neurophysio/. 67, 161>1632.

The Basal Ganglia Matsumura, M., Tremblay,L., Richard, H. and Filion, M. (1995) Activity of pallidal neurons in the monkey during dyskinesia induced by injection of bicucullinein tbe external pallidum. Neuroscience 65, 59-70.

Matthews,P. B. C. (1991)The human stretch reflexand the motor cortex. Trends Neurosci. 14, 87–91. Mauritz, K.-H. and Wise,S. P. (1986)Premotorcortexof the rhesus monkey: neuronal activity in anticipation of predictable environmentalevents. Exp. Brain Res. 61, 229–244. McGeer,P. L., McGeer,E. G., Scherer,U. and Singh,K. (1977)A glutamatergiccorticostriatalpathway?.Brain Res. 128,36%373. Mettler, F. A. (1945)Effectsof bilateral simultaneoussubcortical lesionsin the primate. J, Neuropafh. .Exp. Neurol.4, 99-122. Middleton,F. A. and Strick, P. L. (1994)Anatomicalevidencefor cerebella and basal ganglia involvementin higher cognitive function. .Scierrce266, 458461. Miller, W. C. and DeLong, M. R. (1988) Parkinsonian symptomatology.An anatomicaland physiologicalanalysis.Ann. N. Y. Acad. Sci. 515, 287-302.

Mink, J. W. and Thach,W. T. (1991a)Basalgangliamotor control. 1. Nonexclusiverelation of pallidal dischargeto five movement modes, J. Neurophysiol. 65, 273–300. Mink, J. W. and Thach,W. T. (1991b)Basalgangliamotor control. II. Late pallidal timing relative to movement onset and inconsistent prdlidal coding of movement parameters. J. Neurophysiol. 65, 301–329.

Mink, J. W. and Thach,W. T. (1991c)Basalgangliamotor control. 111.Pallidalablation:normalreactiontime,musclecocontraction, and slow movement.J. Neurophysiol. 65, 33&351. Mink, J. W. and Thach,W. T. (1993)Basalgangliaintrinsiccircuits and their role in behavior.Curr. Opirr. Neurobio/. 3, 950-957. Mitchell,1.J., Clarke,C. E., Boyce,S., Robertson,R. G., Peggs,D., Sambrook, M. A. and Crossman, A. R. (1989a) Neural mechanisms underlying Parkinsonian symptoms based upon regional uptake of 2-deoxyglucosein monkeys exposed to l-methyl-4-phenyl-1,2,3,6-tetrahydropyridine. Neuroscience32, 213-226. Mitchell,1.J., Jackson,A., Sambrook,M. A. and Crossman,A. R. (1989b) The role of the subthalamic nucleus in experimental chorea. Evidencefrom 2-deoxyglucosemetabolic mapping and horseradishperoxidasetracing studies. Brairr 112, 1533-1548. Mitchell, I. J., Sambrook, M. A. and Crossman, A. R. (1985) Subcorticalchangesin the regionaluptake of [’H]-2-deoxyglucose in the brain of the monkey during experimentalchoreiform dyskinesiaelicited by injection of a gamma-aminobutyricacid antagonistinto the subthalamicnucleus. Brain 108,405+22. Mitchell,S. J., Richardson,R. T., Baker,F. H. and DeLong,M. R. (1987)The primate globus pallidus:neuronal activity related to direction of movement.Exp. Brain Res. 68, 491-505. Montgomery,E. B. and Buchholz,S. R. (1991)The striatum and motor cortex in motor initiation and execution.Brain Res. 549, 222-229.

Morris, R. G. M., Kandel, E. R. and Squire, L. R. (1988)The neuroscienceof learningand memory:cells, neural circuits and behaviour. Trends Neurosci. 11, 125-127. Mushiake,H., Inase, M. and Tanji, J. (1990)Selectiveencodingof motor sequencein the supplementarymotor area of the monkey cerebralcortex. .Exp. Brain Res. 82, 208-210. Mushiake, H. and Strick, P. L. (1995a) Cerebella and pallidal activityduring instructeddelayperiods.Soc. Neurosci.Absfr.21, 411. Mushiake, H. and Strick, P. L. (1995b)Pallidal neuron activity during sequential arm movements.J. Neurophysiol. 74, 2754 2758.

Nakanishi,H., Kita, H, and Kitai, S. T. (1987)Intracellularstudy of rat substantialnigra pars reticulateneuronsin an in uitro slice preparation: electrical membrane properties and response characteristicsto subthalamicstimulation.Brain Res. 437, 45–55. Nakanishi,H., Kita, H. and Kitai, S. T. (1991)Intracellularstudy of rat entopeduncular nucleus neurons in an in uilro slice preparation:responseto subthalamicstimulation.Bmirr Res. 549, 285-291.

Nakashima, K., Rothwell, J. C., Day, B. L., Thompson, P. D., Shannon, K. and Marsden, C. D. (1989)Reciprocalinhibition betweenforearmmusclesin patientswithwriter’scrampand other occupationalcramps,symptomatichemidystoniaand hemiparesis due to stroke. Brain 112, 681-697.

423

Nambu, A., Acedia,S. and Jinnai, K, (1990)Dischargeof pallidal neuronswith input from variouscorticalareas during movement in the monkey. Brain Res, 519, 183–191. Nauta, W. J. H. and Mehler, W. H. (1966) Projections of the Ientiforrnnucleusin the monkey. Brain Res. 1, 3-42, Neafaey,E. J., Hull, C. D. and Buchwald,N. A, (1978)Preparation for movementin the cat, 11.Unit activityin the basalgangliaand thalamus. Elecmoenceph. Clin. Neurophysiol, 44, 714-723. Noth, J., Friedemann,H. 1., Podoll, K. and Lange, H, W. (1983) Absenceof long-latencyreflexesto imposedfingerdisplacements in patientswith Huntington’sdisease.Neurosci. L.er/. 35, 97–100, Noth, J., Podoll, K. and Friedemann, H.-H. (1985) Long-loop reflexesin small hand musclesstudiedin normal subjectsand in patients with Huntington’sdisease. Brain 10S,6>80. Oberg, R. G. E. and Divac, 1. (1979)Cognitivefunctionsof the neostriatum.In: The Neosrriarurn, pp. 291–314.Eds. I. Divacand R. G. E. Oberg. Pergamon:Oxford. Oertel,W. H. and Mugnaini,E. (1984)Immunocytochemical studies of GABAergicneuronsin rat basal gangliaand their relationsto other neuronal systems.Neurwci. Letr. 47, 233–238. Panizza, M., Lelli, S., Nilsson, J. and Hallett, M. (1990)H-reflex recoverycurve and reciprocalinhibition of H-reflexin different kinds of dystonia.Neurology40, 824828. Pare, D., Curro’Dossi,R. and Steriade,M. (1990)Neuronalbasisof the parkinsonianrestingtremor:a hypothesisand its implications for treatment, Neuroscience35, 217-226. Parent, A., Bouchard,C, and Smith, Y. (1984)The striatopallidal and striatonigralprojections:twodistinctfibersystemsin primate. Brain Res. 303, 385-390.

Parent, A. and De Bellefeuille,L. (1982)Organizationof efferent projectionsfromthe internalsegmentofglobuspallidusin primate as revealedbye fluorescenceretrograde labelingmethod. Brain Res. 245, 201-213.

Parent, A. and Hazrati, L.-N. (1993) Anatomical aspects of informationprocessingin primatebasalganglia. Trends Neurosci. 16, 111-116. Parent, A. and Smith, Y. (1987a) Differential dopaminergic innervationof the two pallidal segmentsin the squirrelmonkey (Sairrririsciureus).Brain Res. 426, 397400. Parent, A. and Smith, Y. (1987b) Organization of efferent projectionsof the e+bthalamicnucleus in the squirrel monkey as revealed by retrograde labeling methods. Brain Res. 436, 296310.

Parent, A., Smith, Y., Filion, M. and Dumas, J. (1989)Distinct afTerentsto internaland externalpallidalsegmentsin the squirrel monkey. Newarci. .Letl. %, 10144. Park, M. R., Lighthall,J. W. and Kitai, S. T. (1980)Recurrent inhibitionin the rat neostriatum. Brain Res. 194,359-369. Parthasarthy, H. B., Schall, J. D. and Graybiel, A. M. (1992) Distributedbut convergentorderingof corticostriatalprojections: analysisof the frontaleyefieldand the supplementaryeyefieldin the macaquemonkey. J. Neurosci. 12, 44684488. Penney,J. B, and Young,A. B. (1981)GABAas the pallidothalamic neurotransmitter:Implicationsfor basal ganglia function. Brerin Res. 207, 195-199.

Penney,J. B.and Young,A. B.(1983)Speculationson the functional anatomyof basalgangliadisorders.Ann. Rev. Neurosci. 6, 73–94. Penny,G. R., Afsharpour,S. and Kitai, S, T. (1986)The glutamate decarboxylase-,Ieucineenkephalin-,methionineenkephalin-and substanceP-immunoreactiveneuronsin the neostriatumof the rat and cat: evidencefor partial populationoverlap. Neuroscience 17, 1011-1045. Percheron,G. and Filion, M, (1991)Parallelprocessingin the basal ganglia:up to a point. Trends Nerowsci.14, 5S56. Percheron,G., Yehrik,J. and Francois,C. (1984)A Golgianalysis of the primate globus pallidus. 111.Spatial organizationof the striato-pallidalcomplex.J. Comp. Neurol. 227, 214227. Phelps, P. E., Houser, C. R. and Vaughn, J. E. (1985) [mmunocytochemicallocalization of choline acetyltransferase within the rat neostriatum: A correlated light and electron microscopicstudy of cholinergicneuronsand synapses.J. Comp. Neurol. 238, 28&307. Precht, W. and Yoshida, M. (1971)Blockageof caudate-evoked inhibitionof neuronsin the substantialnigra by picrotoxin.Brain Res. 32, 229-233.

Rafal,R. D., Inhoff,A. W., Friedman,J. H. and Bernstein,E. (1987) Programming and execution of sequential movements in

424

J. W. Mink

Parkinson’s disease. J. Neurol. Neurosurg, Psyc/riafr. 50, 1267-1273,

Rea, G, L. and Ebner, T. J. (1991)Quantitativeassessmentof the effectof basal ganglia lesionson the stretch reflexin primates. Stereotacf. Funct. Neurosurg. 56, 44-65.

Reiner,A., Albin,R, L., Anderson,K. D., D’Amato,C. J., Penney, J. B. and Young, A. B. (1988) Differential loss of striatal projection neurons in Huntington disease, Proceedings of fhe National Academy of Sciences of the Unifed Sfafes of America 85, 5733-5737.

Ribak, C. E,, Vaughn, J. E. and Roberts, E, (1979)The GABA neurons and their axon terminals in rat corpus striatum as demonstratedby GAD immunocytochemistry, J. Crwnp. Neuro/. 187,261-283.

Richter, R. (1945)Degenerationof the basal ganglia in monkeys from chronic carbon disulphidepoisoning. J. Newwpafh. Exp. Neur-ol.4, 324353.

Rinvik,E., Grofova,1.and Ottersen,O. P. (1976)Demonstrationof nigrotcctal and nigroreticularprojectionsin the cat”by axonal transport of proteins. Brain Res. 112, 388–394, Rinvik,E. and Otteraen,O. P. (1993)Terminalsofsubthalamonigral fibres are enriched with glutamate-like immunoreactivity:an electronmicroscopic,immunogoldanalysisin the cat. J. Chem. Neuroarraf. 6, 19–30.

Robertson,C. and Flowers,K. A. (1990)Motor set in Parkinson’s disease.J. Neurol. Neurosurg, Psychiafr. 53, 583–592, Robertson,R. G., Farmery,S. M., Sambrook,M. A. and Crossman, A. R. (1989)Dyskinesiain the primate followinginjectionof an excitatoryamino acid antagonistinto the medialsegmentof the globus pallidus. Brain Res. 476, 317–322, Robledo, P. and Feger, J. (1990) Excitatory influence of rat subthalamicnucleus to substantialnigra pars reticulate and the pallidalcomplex:electrophysiological data. Brain Res. 518,47-54. Rolls,E. T., Thorpe,S. J. and Maddison,S. P. (1983)Responsesof striatal neuronsin the behavingmonkey. 1, Head of the caudate nucleus. Behau. Brain Res. 7, 179–210. Romo, R., Scarnati,E. and Schultz,W. (1992)Roleof primatebasal ganglia and frontal cortex in the internal generation of movements.11.Movement-relatedactivityin theanteriorstriatum.

dopamine antagonists into the prefrontal cortex of rhesus monkeysperformingand ocrdomotordelayed responsetask. J. Neurophysiol. 71, 515-528.

Sehlag,J. and Schlag-Rey,M. (1987)Evidencefor a supplementary eye field.J. Neurophysio/, 57, 179-200. Schneider,J. and Lidsky, T. (1981)Processingof somatoscrrsory information in striatum of behavingcats. .J. Neurophysio/. 45, 841–851,

Schneider,J. S., Unguez,G., Yuwiler,A., Berg,S. C. and Markham, C. H. (1988)Deficitsin operanthehaviourin monkeystreatedwith N-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), Brain 111, 126>1285. Schultz,W. (1986)Activityof pars reticulate neurons of monkey substantialnigrain relationto motor, sensory,and complexevents. J. Neurophysiol. 55, 660-677.

Schultz,W., Apicella,P. and Ljunglxrg, T. (1993)Responsesof monkey dopamine neurons to reward and conditioned stimuli during successivesteps of learning a delayed response task. J. Neurosci.13, 90W913, Schu[tz,W. and Romo, R. (1990)Dopamineneuronsof the monkey midbrain: Contingencies of responses to stimuli eliciting immediatebehavioralreactions. J. Neurophysiol. 63, 607+24. Schultz,W, and Romo, R. (1992)Roleof primatebasalgangliaand frontal cortex in the internal generation of movements, 1. Preparatoryactivityin the anterior striatum. Exp. Brain Res. 91, 363-384.

Schultz,W., Romo, R., Ljungberg,T., Mirenowicz,J., Hollerman, J. R. and Dickinson,A. (1995)Reward-relatedsignalscarriedby dopamineneurons. In: Models of Information Processing in fhe Basal Gang/ia, pp. 233-249. Eds, J, C. Houk, J. L. Davisand D, G. Beiser,MIT Press:Cambridge, Schultz,W., Studer, A,, Jonsson,G., Sundstrom,E. and Mefford, I. (1985)Deficitsin behavioralinitiationand executionprncesses in monkeyswith l-methyl-4-pheny l-l,2,3,6-tetrahydropyridine-induced parkinsonism.Neurosci. Lett. 59, 22$232, Schultz,W., Studer,A., Romo, R., Sundstrom,E., Jonsson,G. and Scarnati,E. (1989)Deficitsin reactionstimesand movementtimes as correlates of hypokinesiain monkeys with MPTP-induced striatal dopaminedepletion.J. Neurophysio/. 61, 651-668. Exp. Brain Res. 91, 385395. Selemon, L. D. and Goldman-Rakic,P. S. (1985) Longitudinal Romo, R. and Schultz,W. (1990)Dopamineneuronsof the monkey topographyand interdigitationofcorticostriatalprojectionsin the midbrain: Contingenciesof responses to active touch during rhesusmonkey.J. Neurosci. 5, 77&794. self-initiatedarm movements.J. Neurophysiol. 63, 592–606. Sibley, D. R. and Monsma, F. J. (1992) Molecular biology of Romo, R. and Schultz,W. (1992)Role of the primatebasal ganglia dopaminereceptors. Trends Pharm, .Sci. 13, 61-69. and frontal cortex in the internal generationof movements.111, Smith,A. D. and Bolam,J. P. (1990)Theneura1networkof the basal Neuronal activityin the supplementarymotor area. Exp. Brain ganglia as revealed by the study of synaptic connections of Res. 91, 396407. identifiedneurones, Trends in Neuroscierrces13, 259-265. Rothwell,J. C., Obese, J. A., Traub, M. M. and Marsden, C. D. Smith,Y., Hazrati, L.-N. and Parent, A. (1990)Efferentprojections (1983)The behaviorof the long-latencystretch reflexin patients of the subthalamicnucleusin the squirrelmonkeyas studiedby with Parkinson’sdisease, J. Neurol, Neurosurg. Psychiafr. 46, the PHA-L antergogradetracing method. J. Cwrrp. Neurol. 294, 30&323. 35-44. Rouiller, E. M., Liang, F., Babalian, A., Moret, V. and Strick, P. L., Hoover,J. E. and Mushiake,H. (1993)Evidencefor Weisendanger, M. (1994) Cerebellothalamocorticaland pal“output channels”in the basal gangliaand cerebellum.In: Role of the cerebellum and basal ganglia in uolunfary movement, Iidothalamocorticalprojectionsto the primaryand supplementary motor cortical areas: a multiple tracing study in macaque Pp. 171-182.Eds. N. Mano, I. Hamada and M. R, DeLong. ElsevierSciencePublishers:Amsterdam. monkeys.J. Comp. Neurol. 345, 185–213. Rouzaire-Dubois,B.,Hammond,C., Hamon,B.and Feger,J. (1980) Surmeier,D. J., Foehring, R., Stefani, A. and Kitai, S. T. (1991) Pharmacologicalblockadeof the globuspallidus-inducedinhibiDevelopmental expression of a slowly-inactivatingvoltage dependentpotassiumcurrent in rat neostriatalneurons. Neurosci. tory response of subthalamiccells in the rat. Brain Res. 200, 321-329.

Left, 112, 4146.

Surmeier,D. J., Reiner,A., Levine,M. S. and Ariano,M. A. (1993) Are neostriataldopaminereceptorsco-lncalized?.TrendsNeurosci. 16, 299-305. Tanji,J. (1994)Thesupplementarymotorareain thecerebralcortex. Neuroscience 21,429-440. Neurosci.Res. 19, 251-268. Roy,E. A., Saint-Cyr,J., Taylor,A.andLang,A.(1993)Movement sequencing disordersin Parkinson’s disease.Iret. .f. Neurosci. 73, Tatton, W. G. and Lee, R. G. (1975) Evidencefor abnormal long-loopreflexesin rigid Parkirisonianpatients. Brain Res. 100, 183-194. 671-676. Rye, D. B., Lee, H. J., Saper, C. B. and Wainer, B. H. (1988) Medrdlaryand spinalefferentsnf the pedunculopontinetegmental Tatton, W. G., Redingham,W., Verrier,M. C. and Blair, R. D. G. (1984)Characteristicalterations in responsesto imposed wrist nucleusand adjacmrtmesopontinetegrnentumin the rat. J. Conrp. displacementsin Parkinsonianrigidityand dystoniamusculorum Neural. 269, 31$341. deformans.Can. J. Neuro/. Sci. 11,281-287. Sadikot, A. F., Parent, A. and Francois, C. (1992) Efferent connections of the centromedian and parafaacicularthalamic Thach, W. T,, Goodkin, H, P. and Keating, J. G. (1992) The cerebellumand the adaptivecoordinationof movement.Ann. Reo. nuclei in the squirrel monkey: A PHA-L study of subcortical Neurosci.15, 403-442. projections.J. Comp. Neuro/. 315,137-159. Sawaguchi, T. and Goldman-Rakic, P. S. (1994) The role of Thach,W.T., Mink,J. W., Goodkin,H. P. and Keating,J. G. (1993) Combiningversusgating mntor programs:Differentialroles for D1-dopaminereceptorsin working memory:local injectionsof Rouzaire-Dubois,B. and Scarnati,E. (1987)Pharmacologicalstudy of the cortical-inducedexcitationof subthalamicnucleusneurons in the rat: evidencefor aminoacidsas putativeneurotransmitters.

The Basal Ganglia

425

primate subthalamicnucleus. I Functional properties in intact animals. J. Neurophysiol. 72, 494-506. Hamada and M, R. DeLong. Elsevier Science Publishers: Wichmann, T., Bergman, H, and DeLong, M, R. (1994b)The primatesubthalamicnucleus.111.Changesin motor behaviorand Amsterdam. neuronalactivityin the internalpalliduminducedby subthaiamic Thompson,P. D., Berardelli,A., Rothwell,J. C., Day, B. L., Dick, inactivationin the MPTPmodelof Parkinsonism.J. Neurop/r~,sio/. J. P. R., Benecke,R. and Marsden,C. D. (1988)The coexistence 72, 521-530. of bradykinesia and chorea in Huntington’s disease and its implicationsfor theories of basal gangliacontrol of movement. Wickens,J. and Kotter, R. (1995)Cellularmodelsof reinforcement. In: Models af Information Processing in the Basal Ganglia, Brrzirr111,223-244. Tipton, K. F. and Singer, T. P. (1993) Advances in our PP. 187–214. Eds. J. C. Houk, J. L. Davisand D. G. Beiser.MIT Press:Cambridge. understandingof the mechanismsof the neurotoxicityof MPTP Wilson,C. J. (1995)Thecontributionof corticalneuronsto thefiring and related compounds.J. Neurocherrr.61, 1191–1206. pattern of striatal spiny neurons. [n: Models of Irrforwration Tokuno, H., Kimura, M. and Tanji, J. (1992)Pallidal inputs to Processing in the Baser/Garrg/ia, pp. 29-50. Eds.J. C. Houk, J. L. thalamocorticalneurons projectingto the supplementarymotor Davis and D. G. Beiser.MIT Press:Cambridge. area: an anterogradeand retrogradedouble labelingstudy in the Wilson, C. J., Chang, H. T. and Kitai, S. T. (1982)Origins of macaquemonkey. Exp. Brain Res. 90, 635+38. Tremblay, L., Filion, M. and Bedard, P. J. (1989) Responsesof post-synapticpotentialsevokedin identifiedneostriatal neurons by stimulationin substantialnigra. Exp. Brain Res. 45, 157-167. pallidal neurons to striatal stimulation in monkeys with Wilson,C. J. and Groves,P. M. (1980)Fine structureand synaptic MPTP-induceparkinsonism.Brain Res. 498, 17-33. connectionsof the commonspinyneuron of the rat neostriatum: Trouche, E., Beaubaton, D., Amato, G., Legallet, E. and A study employing intracellular injection of horseradish Zenatti, A. (1979)The role of the internal pallidal segmenton peroxidase.J. Comp. Neurol. 194, 599+14. the execution of a goal directed movement. Brain Res. 175, Wilson, C. J., Kita, H. and Kawaguchi,Y. (1989) GABAergic 362-365. interneurons, rather than spiny cell axon collaterals, are Trouche,E., Beaubaton,D., Amato, G., Viallet,F. and Legallet,E. responsiblefor the IPSP responses to afferent stimulation in (1984)Changesin reactiontime after pallidal or nigral lesionin neostriatal spiny neurons. WC. Neurosci. Abstr. 15, 907. the monkey. Ado. Neurol, 40, 29–38. Ueki, A. (1983) The mode of nigro-thalamic transmission Wilson, S. A. K. (1912) ProgressiveIenticular degeneration:A familialnervous systemdiseaseassociatedwith cirrhosisof the investigatedwithintracellularrecordingin thecat. Exp. Brain Res. liver. Brain 34, 295-507. 49, 1l&124. Ungerstedt, U. (1971)Adipsia and aphagia after 6-hydroxydopa- Wilson, S. A. K. (1928) Modern Probkms irr Neurology. Arnold: London. mineinduceddegenerationof the nigro-striataldopaminesystem. Wing, A. M. (1988)A comparisonof the rate of pinch grip force Acta. Physiol. Scand. 82, 92-122. increasesand decreasesin Parkinsonianbradykinesia.Neuropsyvan den Bercken,J. H. L. and Cools, A. R. (1982)Evidencefor a chologia 26, 479482. role of the caudate nucleus in the sequential organization of Yamamoto,T., Hassler,R., Huber, C., Wagner,A. and Sasaki,K. behaviour. Behav. Brain Res. 4, 319-337. (1983)Eleetrophysiologicalstudies on the pallido- and cerebelvon Krosigk,M., Smith, Y., Bolam,J, P. and Smith, A. D. (1992) Iothalamicprojectionsin the squirrelmonkey.Exp. Brain Res. 51, Synapticorganizationof GABAergicinputsfromthe striatumand 77–87. the globus pallidus onto neurons in the substantialnigra and retrorubral field which project to the medullary reticular Yelnik, J. and Percheron, G. (1979) Subthalamic neurons in primates:A quantitativeand comparativeanalysis.Neuroscience formation. Neuroscierrce50, 531-549. 4, 1717-1743. Vreeling,F, W,, Verhey,F. R. J., Houx, P. J. and Jones, J. (1993) Primitive reflexesin Parkinson’sdisease.J. Neurol. Neurosurg. Yelnik,J., Percheron,G. and Francois,C. (1984)A Golgi analysis of the primate globuspallidus.11.Quantitativemorphologyand Psychiatr. 56, 1323–1326. spatial organization of the dendritic arborizations. J. Ccmrp. Walker,R, H., Arbuthnott, G. W., Baughman,R. W. and Graybiel, Neurol. 227, 20&213. A, M. (1993)Dendriticdomainsof mediumspinyneuronsin the Yeterian,E. H. and Pandya,D. N. (1993)Striatalconnectionsof the primate striatum: relationshipsto striosomalborders. J. Comp. parietalassociationcorticesin rhesusmonkeys.J. Comp. Neurol. Neurol. 337, 61&528. 332, 175-197. Watts, R, L. and Mandir, A. S. (1992)The role of motor cortexin the pathophysiologyof voluntary movementdeficitsassociated Yeterian, E. H. and Van Hoesen, G. W. (1978) Cortico-striate projectionsin the rhesus monkey: The organizationof certain with Parkinsonism.Neurol. C/in. 10,451-469. cortico-caudateconnections.Brain Res. 139, 43–63. Wichmann, T., Bergman, H. and DeLong, M, R. (1994a) The cerebellumand basal ganglia.In: Role qf~he cerebellum and basal ganglia in uo[untary rnouement, pp. 235-245. Eds. N. Mano, I.