Diagram of overall cerebellar organization. Transverse cortical folds comprise folia, lobules, and lobes. Shown is longitudinal pattern of projection of cortical Purkinje cells onto deep nuclei and their targets. Mossy fiber inputs often branch to reach both nuclei and cortex. Also shown is origin of mossy fibers supplying different subdivisions and modalities of information that they are likely to carry. Pontine and medullary tegmental reticular nuclei supply all of cerebellum with mossy fibers (not shown). VL, ventrolateral (nucleus of thalamus); lateral (L), intermediate (I), medial (M) cerebellar cortex; D, dentate nucleus; IP, interposed nucleus; F, fastigial nucleus.

Surmises about major cerebellar connections and functions. A: 1974 scheme showing proposed roles of several brain structures in movement. Short dashed lines at A and B represent lesions described in text. It is proposed that basal ganglia and cerebellar hemisphere are involved with association cortex in programming volitional movements. At time that motor command descends to motoneurons, engaging movement, pars intermedia updates intended movement based on motor command and somatosensory description of limb position and velocity on which movement is to be superimposed. Follow‐up correction can be performed by motor cortex when cerebellar hemisphere and pars intermedia do not effectively perform their functions. B: 1981 scheme showing intended movements seen with reference to cerebellar circuit. It is suggested that phasic cerebellar control involves triggering small programs such as “start” and “stop,” the subprograms of which are stored in various parts of circuit. Supplementary motor area (area 6) is viewed as portal for “higher instructions.” Lateral and intermediate cerebellum both connect to areas 4 and 6, but through unknown links to the latter. They are thought to trigger movement “start” and “stop” commands. No statement is made about medial cerebellum because its actions, such as coordinating stance, have been least studied by behavioral methods.

Oscarsson's scheme of cerebellum as a comparator (1973). Some paths between cerebral motor cortex, anterior lobe, and lower motor centers with an interpretation of function of these paths. Anterior lobe assumed to correct errors in motor activity elicited from cerebral cortex and carried out by command signals through pyramidal and extrapyramidal paths. Command signals assumed to be monitored by anterior lobe through paths relayed in inferior olive and pontine and reticular nuclei. Spinocerebellar paths assumed to serve as feedback channels that monitor activity in lower motor centers and evolving movement. Activity in lower motor centers would be compared with original command signals in anterior lobe and in precerebellar nuclei (cf. Fig. 5). On basis of this comparison corrections might already have begun before initiation of the movement. Evolving movement would be compared with original commands and with activity in lower motor centers and results of comparison used for corrections during movement. Direct mossy fiber paths: DSCT, proprioceptive component of dorsal spinocerebellar tract; CCT, proprioceptive component of cuneocerebellar tract; VSCT, ventral spinocerebellar tract; RSCT, rostral spinocerebellar tract. Indirect mossy fiber path: LRN‐SRCP, dorsal spinoolivocerebellar path relayed through lateral reticular nucleus. Climbing fiber paths: DF‐SOCP, dorsal spinoolivocerebellar path; DLF‐SOCP, dorsolateral spinoolivocerebellar path; VF‐SOCP, ventral spinoolivocerebellar path; LF‐CF‐SCP, lateral climbing fiber‐spinocerebellar path; VF‐CF‐SCP, ventral climbing fiber‐spinocerebellar path.

Intended, simple, self‐terminated movements. A: fast flexions made by normal human subjects of elbow (left) and top joint of thumb (right). Note triphasic EMG pattern in these “skilled ballistic” movements made “as fast and as accurately as possible.” Records of single traces from above down: EMG activity in agonist

The “cerebellar circuit” drawn with reference to intended movements. A: main connections; it is suggested that this distributed system may implement intended simple (and hence compound) movements that fit task context. Diagram is partial and oversimplified, nuclear groups combined, and connections of particular parts omitted. Parallel efferent lines descend from cerebrum to spinal cord and parallel afferent lines from cord to cerebrum and to cerebellum (hatched and striped arrows), whose output (cerebellum, stippled) nuclei include fastigial, interposed, and dentate. Precerebellar nuclei (pontine and lateral reticular nuclei and inferior olive) receive input from parietal (area 5), sensory (areas 1 and 2), and motor cortex (area 4). “Higher” inputs convey intent to supplementary motor cortex (area 6: open arrow). Outputs from motor cortex are indicated by black arrows, corticocortical loops by small arrows. Route of implementation from cerebellum to cerebral cortex is drawn through “motor” thalamus, simply called VL. B: selection of precentral cell assemblies could trade speed for accuracy to fit movement to intent. Cerebellothalamic projections may selectively facilitate precentral neurons with appropriate response capabilities to peripheral inputs to implement movements of various speeds. Arrows (with dotted adjusters) indicate changeable effectiveness of connections. Place of this scheme in cerebellar circuit shown in A. VL, ventrolateral thalamus; PTNs, pyramidal tract neurons.

Compound movements of normal human subjects left) and patients with cerebellar deficits (right). Quantitative tests of functional stretch responses (FSRs) in upper 2 rows (A, B, E, F) and of vestibular function in lower 2 rows (C, D, G, H). Left column: performance of normal subjects in each of 4 tests categories: A, induced sway; B, direct angle rotation; C, induced sway with stabilized angles; and D, induced sway with stabilized angles and no vision. EMGs are from gastrocnemius (G); hamstrings (H); tibialis anterior (T); and quadriceps (Q). M is moment exerted by ankle musculature upon platform, and θ is body sway angle. Graphs at right show average performance ± SD for normals in each test category. Right column: dysfunctional responses in each of 4 test categories (E‐H). Raw data are presented in same form as used in A‐D. Except where indicated, however, only activity of gastrocnemius muscle is shown. At right, minimum criteria for classifying response as abnormal are given. K: correlation between clinical grade and platform tests. Of 15 patients, 4 were excluded because of incomplete data and extensive CNS lesions. Abnormality was limited to 1 mode if only FSR or vestibular categories were involved but not to both. Total number of categories among FSR adaptation, FSR pattern, vestibular static, and vestibular dynamic also shown.

Rhythm of rapid alternating movements slowed by cerebellar dysfunction. A: “rapid” alternate supination and pronation of right (R) and left (L) arms in case of gunshot wound on left side of cerebellum. Time calibration applies to both traces. B: rhythm of alternating ballistic movements can be programmed. It is maintained after unexpected changes of position of mechanical barriers stopping elbow movements of normal monkey (setup in Fig. 10 B). Records of single traces. Distances between barriers from left to right: 90°, 60°, 60°, and 25°. Calibration 90° for horizontal lines that indicate barrier positions for flexions at top and extensions at bottom. C: rhythm of alternate ballistic movements is prolonged by ipsilateral dentate cooling. Position, velocity, and EMG during control (above) and cooling of dentate nucleus (below); single traces. Vertical broken lines and arrows indicate moment of impact on mechanical barriers (“stops”), followed by recoil bounce. Note prolongation of holding and of EMG by 0.1–0.2 s. Same animal as in B. Time axis expanded nearly 10‐fold from B. Velocity calibration: 100°/s.

Delayed movement onset and slowed acceleration caused by cerebellar injury from a case of injury to right lateral lobe. Patient was asked to grasp simultaneously with his two hands against two springs of equal strength: 1 and 1′ represent simultaneous ordinates; A and A', lines traced on rapidly revolving drum. Drum was allowed to complete one revolution and a signal to relax was then given; 2 and 2′ represent simultaneous ordinates; B and B', tracings of relaxation. Time by a tuning fork of 128 vibrations/s. Time markers 0.1 s.

Inhibition of antagonist muscle before normal movement onset is lost after cerebellar injury. Left: inhibition of tonic triceps activity before 4 successive fast elbow flexions of normal human subject. Surface‐recorded EMGs of first biceps burst (upper 4 traces) and triceps (lower 4 traces) for same movements. Triceps fired tonically against a weight of about 2 kg. Time calibration: 20 ms. Right patient with mild ataxia exhibits loss of triceps inhibition in 8 successive trials, as at left. Note that triceps activity usually ceases the moment biceps begins or shortly thereafter but never significantly before, as it normally should.

Onset of intended movements and related precentral discharges are both equally delayed by cerebellar dysfunction. A: three possible schemes for initiation of prompt arm movements that are not necessarily mutually exclusive. Left, sequences after “Move!” signal; right, expected time relations between discharge of movement‐related precentral neuron and movement onset (triangles). Expected results shown diagrammatically for control conditions above horizontal lines, and during cerebellar cooling below. Precentral discharge follows visual “Go!” signal (vertical arrow). The three schemes predict three different discharge patterns during cerebellar nuclear cooling. Tonic precentral discharge levels would decrease in 1, but not in 2 and 3. Timing of discharge in relation to movement onset (triangles) differentiates between 2 and 3: precentral neural onset times would remain unchanged in 2 but would be delayed in 3. B: monkey in experimental chair with elbow supported on pivot of manipulandum and hand grasping handle, torque motor beneath. Juice reward tube in front of mouth; head holder and microdrive not shown. Opaque plate blocking monkey's view of task area indicated by dotted lines. Monkey has learned to terminate movements in mechanically undetectable target zones, such as the one drawn (hatched). C: delayed onset of movements and of discharge of movement‐related precentral neuron during dentate cooling corresponding to scheme 3 and typical of majority sampled. Upper row: discharges during first 25 movements of a trial plotted in raster (dot) format in successive lines, one below the other with respect to “Go!” signal (vertical arrow at time 0 ms). Lower row. discharge during same movements, but trials have been ordered in terms of increasing latency of movement onset (triangles). Vertical line at 200 ms drawn to assist in recognition of delay in onset of precentral discharge and movement by about 0.15 s.

Slow intended movements oscillate near 3 Hz during cerebellar dysfunction. A: tracings of slow flexion and extension of right (R) and left (L) index fingers in a case of right‐sided wound of cerebellum. B: slow elbow movements of monkey (in setup of Fig. 10B) during ipsilateral dentate cooling and 1 min later. Note poorly controlled termination of self‐paced movements and maintenance of position in targets (gray bands) during cooling, as well as overlapping biceps and triceps activity. Single traces. Calibrations: 35°, 200°/s, 1,500°/s². C: oscillations are synchronized with movement onset in averaged records of self‐paced movements from a similar experiment with another monkey (±SD, dots). Note that these changes are reverse of those occurring during motor learning (Fig. 4C). Normalized frequency histograms in bottom row show that discharge frequency of related precentral neuron varied in rhythm with oscillations. Compare with Fig. 13. Calibrations: 25°, 100°/s, 1,000°/s², and 100 impulses/s.

Precise coordination of muscles with movement of their joints is lost during cerebellar dysfunction. A, B: “skilled ballistic movements” in patients with cerebellar ataxia. In A, fast elbow flexion from 52‐yr‐old man with spinocerebellar degeneration. Note, by comparison with normal subject in Fig. 4A, left, prolongations of initial agonist and antagonist bursts of EMG activity. In B, fast thumb flexion shown from 62‐yr‐old man with unilateral cerebellar ataxy due to stroke. By comparison to normal in Fig. 4A, right, note slower movement, inability to hold final position, and prolonged bursts of activity in agonist and antagonist muscles. Single traces. Calibration 25° or 313°/s. C: spontaneous elbow flexion made by cerebellectomized, deafferented monkey. Note, as in A, dispersion of antagonist (triceps) EMG in relation to movement. Discharge of flexion‐related precentral neuron precedes movement onset.

Demonstration of transcortical loop effects and their cerebellar control. Cerebellar dysfunction slows rhythm and depresses amplitudes of “late” precentral responses to limb perturbation. A: plot of averaged self‐paced and self‐terminated flexions (± SD, dots). Monkey's arm was perturbed by torque pulse opposing flexion (at time 0) eliciting synchronized oscillations (same setup as in Fig. 10B). During ipsilateral dentate cooling their rhythm slowed from near 6 Hz to near 3 Hz (cf. Fig. 11C). Note that these changes are reverse of those occurring during motor learning (Fig. 4C). Bottom row shows responses of a precentral neuron as normalized frequency histograms. Note that “early” precentral responses, defined in this initial description as having latencies of 2–40 ms, increased with a load on the arm, as “expected” for stretch responses in transcortical loop. “Early” responses unaffected by dentate cooling, but “late” responses depressed in amplitude and their rhythms slowed in cadence with that of movement oscillations. Displays equivalent to Fig. 11C. Calibrations: 25°, 200°/s, 5,000°/s², and 100 impulses/s. B: rhythms of limb oscillations evoked by perturbations and those of precentral “late” responses remain linked during ipsilateral dentate dysfunction. Arrow indicates slowing from control values (warm, ○); plotted for 11 neurons from experiments as in A (± SD, bars).

Demonstration of predictive damping by transcortical loop and its cerebellar control. Panel 1: cerebellar dysfunction by nuclear cooling depresses second, but not first, precentral response to perturbation of monkey's arm from an intended held position. Normalized frequency histograms of neuron for first and second responses, with latencies of 20–50 ms (black) and 50–100 ms (hatched), respectively. These two responses had not yet been distinguished from one another in “early” responses shown in Fig. 13A Calibration: 75 impulses/s. Panel 2: depression of second precentral response delays braking of movements. Effect of cerebellar nuclear cooling (IP and dentate) on reciprocal precentral neuron and corresponding position and EMG records. Unit closely related to EMG of biceps muscle. During ipsilateral cooling there is no change in first cortical response (20–50 ms), which could contribute to largely unchanged late stretch‐reflex responses in biceps (left, biceps stretch). However, cooling depresses (50–100 ms) second cortical response (right, triceps stretch) when biceps is antagonist. Therefore, during cooling precentral neural discharge is no longer phase advanced to onset of biceps activity, but now follows biceps stretch, and oscillations that are normally damped during braking now become undamped. Dashed line at 50 ms, drawn for reference. Records represent averages of 25 trials; IPS, impulses per second.

Connections of inferior olive suggest that it could function as a comparator. A: possible functional relation between a sagittal zone in cerebellar cortex, motor center controlled by this zone, and ascending and descending climbing fiber paths to this zone in cat. The “d₁” zone is assumed to control a motor center in lumbar enlargement, which facilitates dynamic γ‐motoneurons (mn) and inhibits effects from flexor reflex afferents (FRA) to α‐motoneurons and primary afferents (primary afferent depolarization, PAD). Part of inferior olive (IO) projecting to the d₁ zone may possibly compare (arrow) descending command signals from red nucleus (RN) with the activity that these signals evoke in lower motor center, which is also under segmental afferent control. On basis of this comparison, d₁ zone might help to integrate motor activity evoked from higher centers and reflex activity. Lower motor center is controlled by rubrobulbospinal path (RBSP), which originates from rostral RN and which is under control from highest motor centers and from d₁ zone through dentate nucleus. After a relay in reticular formation (RF), RBSP descends in contralateral dorsolateral funiculus to lumbar enlargement. Motor center is also controlled by distal cutaneous afferents. Activity in center is monitored by SOCP ascending path through dorsolateral funiculus (DLF‐SOCP), which is formed by ascending axon collaterals from neurons in this center. B: comparator hypothesis of inferior olive. Diagram is centered around functional unit consisting of sagittal zone with its olivary region and cerebellar nucleus and lower motor center controlled by this unit (thick outlines). It is assumed that olivary region monitors commands from higher motor centers, activity these commands evoke in the lower motor center, and the resulting movement. By comparing information from these sources, the olive would detect perturbations that might be used by sagittal zone to send signals of correlation either directly to the lower center (path 1) or to the higher centers (path 2).

Simplified diagram of cerebellar circuitry. mf, Mossy fiber; cf, climbing factor; gr, granule cell; Go, Golgi cell; b, basket cell; s, stellate cell; P, Purkinje cell; n, nuclear cell. White cells are excitatory, black are inhibitory. Diagram shows only what types of cell one type contacts and whether contact is excitatory or inhibitory. Omitted are the recently discovered nucleocortical fibers 332.

A: maintained discharge of Purkinje cell, recorded extracellularly, showing its two different spike potentials: the “simple” (left) and “complex” (right). Slow trace (top) shows their different pattern of discharge, fast traces (bottom 3) their different shape. Positivity is up. B: rapid alternating movement. Discharge of Purkinje cell with monkey quiet (A), during movement of ipsilateral wrist (B), ipsilateral shoulder (C), contralateral wrist (D), and contralateral shoulder (E). Line below unit discharge represents movement of manipulandum lever: for wrist movements, up is flexion and down is extension; for shoulder movements, up is pushed and down is pulled. Simple spike of this unit alters its frequency of discharge in a consistent temporal relationship to movements of the ipsilateral wrist but not to other movements. Complex spike (dotted) occurs in no obvious relation to the movement. C: rapid alternating movement. Discharge of dentate cell with monkey quiet (A), during movement at ipsilateral wrist (B), ipsilateral shoulder (C), contralateral wrist (D), and contralateral shoulder (E). Line below unit discharge represents movement of manipulandum lever: for wrist movements, up is flexion and down is extension; for shoulder movements, up is pushed and down is pulled. Cell discharge was related to movement of ipsilateral wrist, less so to movement of ipsilateral shoulder, and not at all consistently to either movement of contralateral arm.

A: timing in relation to prompt wrist movement. Distribution of time of change of discharge frequency relative to flexion or extension (whichever is earlier) for Purkinje cells compared to distributions for dentate and interposed nuclear cells. Onset of movement, 0 ms. B: timing relative to acoustic startle response. Distributions of latencies of initial changes in nuclear, Purkinje, and triceps EMG activity following sound stimulus at time 0. C: timing in relation to prompt wrist movement. Histograms show distribution of time of change (relative to change of force) of unit discharge in dentate nucleus and arm area motor cortex, and for muscles in arm, shoulder, and trunk. Abscissa is time of change (in ms) before or after change of force (↑); scale is 20 ms per division. Ordinate is number of neural (and EMG) changes; scale is 5 neural changes per division. D: timing relative to light‐triggered movement (“volitional,” left) and to perturbation of holding (segmental reflex and longer loops, right). On abscissa, times of change relative to reference point (arrow) in milliseconds.

Timing relative to prompt wrist movement. Discharge of 3 Purkinje cells (A, B, and C) related to flexor (left) and extensor (right) movements. Large dots represent complex spikes; small dots, simple spikes; each row of dots, discharge during a trial; dashed line, onset of movement. Time scale, 50‐ms intervals.

Relation of cerebellar nuclear neural discharge to body part of moving monkeys. Horizontal sections shown through cerebellar nuclei in three experiments. A, B: filled circles represent those penetrations in which cells whose activities were consistently related to forelimb movement were located; open circles represent penetrations in which cellular activity displayed no apparent consistent relationship with arm movements. Crosses denote electrode tracks in which no successful recordings were made. F, fastigial nucleus; ND, dentate nucleus; NIP, interpositus nucleus. C: filled circles represent those penetrations in which maintained discharge during wrist holding was visibly higher for 1 or more of 8 hold conditions than for others; open circles represent penetrations in which it was not. Figure represents combined results from 2 monkeys. Scale is 1 mm/division. D: each penetration represented by filled circle. Letters represent movement of single body part: F, face; T, thumb; W, wrist; E, elbow; S, shoulder; L, leg. Size of letter represents degree of representation of each movement in penetration.

Different durations of modulation of discharge of single Purkinje cell during movement performances involving different durations of pull and hold phases of lever movement. A: both simple and complex spike discharges (dots) are modulated but in opposite directions. Trace A, EMG activity in flexor digitorum profundus; trace B, EMG of brachialis muscle; trace C, discharges of Purkinje cell; trace D, record of displacement of lever; trace E, sudden, unexpected perturbation of lever‐pulling movement performance imposed during pull phase is indicated by disturbance in displacement record. Averaging responses of neuron to a number of such disturbances indicated that no consistent change in neuronal activity occurred in response to perturbation. B: graph of relationship between duration of increase in Purkinje cell activity illustrated in A and duration of the pull phase of the movement task for a large number of repetitions.

Relation to hold. A: discharge in dentate nucleus of a neuron whose maintained frequency during maintained flexion of wrist (middle of each trace) was over twice that during maintained extension (ends of each trace). Top, onset and duration of light signal to change from one hold to other; middle, discharge of unit; bottom, force exerted by hand on lever (up is flexion, down is extension), B: discharge of interpositus neuron related to wrist hold against torque load. There are 8 periods of position holding by wrist (each under a unique condition) followed by a move. There are 8 rasters of neural discharge; frequencies during holds at left. Top 4 holds under a flexor load; bottom 4 under extensor load. C: relation of discharge of interpositus neuron in B to magnitude of torque load.

Coding of neural discharge during maintained holds. Graphs show degree of fit (Q = 0 − 1) of discharge to each of 3 variables to which neurons might best relate: muscles that were active to maintain the hold, flexion or extension (MPAT); or joint position that was held (JPOS); or anticipation of direction of next move, flexion or extension (DSET). The 3 variables (MPAT, JPOS, DSET) were varied independently in experiment. A: for interpositus neurons, left graph shows better fit to MPAT than to DSET; middle graph exhibits a better fit to MPAT than to JPOS; and right graph shows some relation to JPOS and little relation to DSET. In general, discharge of interpositus neurons was determined mainly by direction of load (as was pattern of muscular activity). There was a joint positional (or muscle length) component for a few neurons but little relation to direction of intended next movement. B: for dentate neurons there was relatively less variation in discharge during holds, giving relatively poorer fits to any of the 3 hold variables. Yet, some neurons discharged in as good relation to joint position (JPOS) or anticipated direction of next move (DSET) as others did to those muscles that were active (MPAT).

Coding of muscle activity (A) and dentate discharge (B) to “volitional” and “reflex” aspects of performance in Tanji‐Evarts paradigm. A: independence of early phase EMG activity from motor responses (A1, B1) corresponding to these rasters. Rasters and average responses based on rectified EMG signals recorded from surface electrodes overlying biceps in human subjects. For both conditions, A and B, a load change imposed by a torque motor stretched biceps (at S). In A, subject was given a prior instruction to oppose load. For B, his instruction was to assist or to move in same direction as load. Early components (25–100 ms) of EMG are similar for both instructions, with volitionally determined differences appearing at 100 ms. B: influence of motor preparation on short‐latency response of a dentate neuron to load changes. Neural response averages (A1, B1), rasters of individual trials (A2, B2), and average displacement traces (A3, B3) of same dentate neuron recorded under two conditions. Load change in A and B moved handle away from monkey. In A, animal was instructed to prepare to push, and in B to prepare to pull. Each line in rasters (A2, B2) represents an individual trial and each dot a single neural discharge. Both rasters and response averages show that the same direction of load change evoked a short‐latency increase in activity when animal was prepared to push and a short‐latency decrease in activity when animal was prepared to pull. Maximum scale in A1 is 512 pulses/s and in B1 256 pulses/s.

Dissociation of 3 variables in a task requiring holds. Diagrammatic representation of task (top) and hypothetical discharge frequencies of muscle or CNS neurons related to 3 aspects of task during hold period (bottom); pattern of muscular activity (MPAT), wrist joint position (JPOS), and direction of intended next light‐triggered movement (DSET). Examples of each type of neural discharge were subsequently found in trained monkey, differentially located. Interpositus neurons all resembled MPAT model, whereas a dentate (or motor cortex) neuron related to any one of all 3 models.

Components of interpositus discharge related to an onset signal (early) and a velocity signal (later). Averaged data for a type I interpositus neuron (ip) during nonballistic elbow flexion (A) and flexion initiated by overcoming small (B) and large (C) resistive forces. Arrows under ip records mark peak accleration, peak velocity, and peak negative accleration. A large inertial mass was added in A. Note that the time relation between changes in discharge rate of neuron and motion parameters remains approximately the same, even though duration of the movements (time‐to‐peak displacement) decreases from A to C. Also, biceps activity (bi) covaries with unit activity. Bin width is 4 ms, and 24 trials are averaged.

Discharge of Purkinje cell showing relation to movement onset but not direction of movement. Simple‐spike frequencies of response‐locked Purkinje cell with quick wrist extension (A) and flexion (B). Histograms, positions (P), and velocities (V) are aligned at R event. Horizontal arrows beneath a and b indicate times for determination of control frequencies for ordinate (simple spikes/s); a and b indicate steady‐state 500 ms from 1.5 s after the R event. Histograms were made of indicated trial numbers. P and V were averaged over 10 trials. Calibrations of P (10°), V (50°/s), and time scale (0.5 s) apply to both A and B.

α‐γ‐Motoneuron dissociation and the α‐γ‐motoneuron relation of motor cortex and cerebellar nuclear neurons. A: EMG of wrist flexor in forearm during slow wrist tracking. Top: schematic hold‐ramp‐hold tracking for extension (left) and flexion (right). Graphs are of EMG intensity (ordinate) versus time (abscissa) during extension (left) and flexion (right). Top graphs show discharge under flexor load; middle graphs, under no load; and bottom graphs, under extensor load. EMG was related to direction of load and to position. B: discharge of one type of precentral cerebral cortex neuron during slow wrist tracking. This neuron shows patterns of discharge under different conditions that are similar to activity of a wrist flexor muscle. C: discharge of second type of precentral cerebral neuron. This neuron shows pattern of bidirectional discharge that is remarkably constant under all conditions. D: discharge of a cerebellar nuclear neuron. This nuclear neuron and all others showed bidirectional discharge like second category of neurons in motor cortex. E: discharge of extensor muscle EMG (upper trace) and unit discharge (lower trace) of neuron in C8 dorsal root ganglion. Unit was identified as primary afferent from spindle within extensor muscle. EMG discharge resembles directional discharge of type I precentral cerebral cortex neurons. By contrast, activity of this spindle afferent and of all others resembled bidirectional discharge of type II precentral cerebral cortex neurons and of all cerebellar nuclear neurons.

Cerebellum in adaptive motor control. A: changes in task performance with introduction of a novel load. Task was to move a handle in horizontal arc by flexing or extending wrist to central position and to try to hold it there despite flexor and extensor loads applied to handle. Each trace starts as load switched to one in opposite direction, displacing handle from the central position for a transient period of about 300 ms. Position traces of handle are shown for successive trials (top to bottom) alternately against flexor and extensor loads. Each flexor trace on left is followed by extensor trace on right. With known loads, position traces were smooth and reproducible from trial to trial (above arrow). When extensor load was increased from known 300 g to novel 450 g (arrow), there were immediate irregularities in position traces during transient and maintained periods, which gradually diminished with further trials (below arrow). For flexor trials there were a few irregularities in transient period only as load switched from novel to known. B: complex (CS) and simple (SS) spike frequency changes for Purkinje cell after change in load. Each dot represents spike potential (SSs, small dots; CSs, large dots); each row of dots represents the discharge during a trial, beginning at change in direction of load. Successive trials represented top to bottom, each flexor trace on left followed by an extensor trace on right. This cell was load‐related with higher SS frequency in maintained period for flexor than for extensor trials. At arrow, known extensor load of 300 g was changed to a novel 450 g while known flexor load of 310 g was kept constant. Before load change (above arrow), there was a low frequency of related CS activity at about 100 ms after start of extensor trials. After load change (below arrow) CS frequency at that time increased greatly and persisted for about 70 trials. There was also an increased CS frequency in the extensor maintained period for about 40 trials. Associated with these transient increases in CS frequency there were decreases in SS frequency that persisted. C: relationship of motor performance and complex‐ and simple‐spike frequencies over multiple trials.