Spatial and Temporal Arrangement of Recurrent Inhibition in the Primate Upper Limb.

Renshaw cells mediate recurrent inhibition between motoneurons within the spinal cord. The function of this circuit is not clear; we previously suggested based on computational modeling that it may cancel oscillations in muscle activity around 10 Hz, thereby reducing physiological tremor. Such tremor is especially problematic for dexterous hand movements, yet knowledge of recurrent inhibitory function is sparse for the control of the primate upper limb, where no direct measurements have been made to date. In this study, we made intracellular penetrations into 89 motoneurons in the cervical enlargement of four terminally anesthetized female macaque monkeys, and recorded recurrent IPSPs in response to antidromic stimulation of motor axons. Recurrent inhibition was strongest to motoneurons innervating shoulder muscles and elbow extensors, weak to wrist and digit extensors, and almost absent to the intrinsic muscles of the hand. Recurrent inhibitory connections often spanned joints, for example from motoneurons innervating wrist and digit muscles to those controlling the shoulder and elbow. Wrist and digit flexor motoneurons sometimes inhibited the corresponding extensors, and vice versa. This complex connectivity presumably reflects the flexible usage of the primate upper limb. Using trains of stimuli to motor nerves timed as a Poisson process and coherence analysis, we also examined the temporal properties of recurrent inhibition. The recurrent feedback loop effectively carried frequencies up to 100 Hz, with a coherence peak around 20 Hz. The coherence phase validated predictions from our previous computational model, supporting the idea that recurrent inhibition may function to reduce tremor.SIGNIFICANCE STATEMENT We present the first direct measurements of recurrent inhibition in primate upper limb motoneurons, revealing that it is more flexibly organized than previous observations in cat. Recurrent inhibitory connections were relatively common between motoneurons controlling muscles that act at different joints, and between flexors and extensors. As in the cat, connections were minimal for motoneurons innervating the most distal intrinsic hand muscles. Empirical data are consistent with previous modeling: temporal properties of the recurrent inhibitory feedback loop are compatible with a role in reducing physiological tremor by suppressing oscillations around 10 Hz.

Extensive Cortical Convergence to Primate Reticulospinal Pathways.

Early evolution of the motor cortex included development of connections to brainstem reticulospinal neurons; these projections persist in primates. In this study, we examined the organization of corticoreticular connections in five macaque monkeys (one male) using both intracellular and extracellular recordings from reticular formation neurons, including identified reticulospinal cells. Synaptic responses to stimulation of different parts of primary motor cortex (M1) and supplementary motor area (SMA) bilaterally were assessed. Widespread short latency excitation, compatible with monosynaptic transmission over fast-conducting pathways, was observed, as well as longer latency responses likely reflecting a mixture of slower monosynaptic and oligosynaptic pathways. There was a high degree of convergence: 56% of reticulospinal cells with input from M1 received projections from M1 in both hemispheres; for SMA, the equivalent figure was even higher (70%). Of reticulospinal neurons with input from the cortex, 78% received projections from both M1 and SMA (regardless of hemisphere); 83% of reticulospinal cells with input from M1 received projections from more than one of the tested M1 sites. This convergence at the single cell level allows reticulospinal neurons to integrate information from across the motor areas of the cortex, taking account of the bilateral motor context. Reticulospinal connections are known to strengthen following damage to the corticospinal tract, such as after stroke, partially contributing to functional recovery. Extensive corticoreticular convergence provides redundancy of control, which may allow the cortex to continue to exploit this descending pathway even after damage to one area.SIGNIFICANCE STATEMENT The reticulospinal tract (RST) provides a parallel pathway for motor control in primates, alongside the more sophisticated corticospinal system. We found extensive convergent inputs to primate reticulospinal cells from primary and supplementary motor cortex bilaterally. These redundant connections could maintain transmission of voluntary commands to the spinal cord after damage (e.g., after stroke or spinal cord injury), possibly assisting recovery of function.

Axonal plasticity underpins the functional recovery following surgical decompression in a rat model of cervical spondylotic myelopathy.

Cervical spondylotic myelopathy (CSM) is the most common spinal cord disorder and a major cause of disability in adults. Improvements following surgical decompression are limited and patients often remain severely disabled. Post mortem studies indicate that CSM is associated with profound axonal loss. However, our understanding of the pathophysiology of CSM remains limited.To investigate the hypothesis that axonal plasticity plays a role in the recovery following surgical decompression, we adopted a novel preclinical model of mild to moderate CSM. Spinal cord compression resulted in significant locomotor deterioration, increased expression of the axonal injury marker APP, and loss of serotonergic fibres. Surgical decompression partially reversed the deficits and attenuated APP expression. Decompression was also associated with axonal sprouting, reflected in the restoration of serotonergic fibres and an increase of GAP43 expression. The re-expression of synaptophysin indicated the restoration of functional synapses following decompression. Promoting axonal plasticity may therefore be a therapeutic strategy for promoting neurological recovery in CSM.

Corticospinal Inputs to Primate Motoneurons Innervating the Forelimb from Two Divisions of Primary Motor Cortex and Area 3a.

Previous anatomical work in primates has suggested that only corticospinal axons originating in caudal primary motor cortex ("new M1") and area 3a make monosynaptic cortico-motoneuronal connections with limb motoneurons. By contrast, the more rostral "old M1" is proposed to control motoneurons disynaptically via spinal interneurons. In six macaque monkeys, we examined the effects from focal stimulation within old and new M1 and area 3a on 135 antidromically identified motoneurons projecting to the upper limb. EPSPs with segmental latency shorter than 1.2 ms were classified as definitively monosynaptic; these were seen only after stimulation within new M1 or at the new M1/3a border (incidence 6.6% and 1.3%, respectively; total n = 27). However, most responses had longer latencies. Using measures of the response facilitation after a second stimulus compared with the first, and the reduction in response latency after a third stimulus compared with the first, we classified these late responses as likely mediated by either long-latency monosynaptic (n = 108) or non-monosynaptic linkages (n = 108). Both old and new M1 generated putative long-latency monosynaptic and non-monosynaptic effects; the majority of responses from area 3a were non-monosynaptic. Both types of responses from new M1 had significantly greater amplitude than those from old M1. We suggest that slowly conducting corticospinal fibers from old M1 generate weak late monosynaptic effects in motoneurons. These may represent a stage in control of primate motoneurons by the cortex intermediate between disynaptic output via an interposed interneuron seen in nonprimates and the fast direct monosynaptic connections present in new M1. SIGNIFICANCE STATEMENT: The corticospinal tract in Old World primates makes monosynaptic connections to motoneurons; previous anatomical work suggests that these connections come only from corticospinal tract (CST) neurons in the subdivision of primary motor cortex within the central sulcus ("new M1") and area 3a. Here, we show using electrophysiology that cortico-motoneuronal connections from fast conducting CST fibers are indeed made exclusively from new M1 and its border with 3a. However, we also show that all parts of M1 and 3a have cortico-motoneuronal connections over more slowly conducting CST axons, as well as exert disynaptic effects on motoneurons via interposed interneurons. Differences between old and new M1 are thus more subtle than previously thought.

Pathways mediating functional recovery.

Following damage to the motor system (e.g., after stroke or spinal cord injury), recovery of upper limb function exploits the multiple pathways which allow motor commands to be sent to the spinal cord. Corticospinal fibers originate from premotor as well as primary motor cortex. While some corticospinal fibers make direct monosynaptic connections to motoneurons, there are also many connections to interneurons which allow control of motoneurons indirectly. Such interneurons may be placed within the cervical enlargement, or more rostrally (propriospinal interneurons). In addition, connections from cortex to the reticular formation in the brainstem allow motor commands to be sent over the reticulospinal tract to these spinal centers. In this review, we consider the relative roles of these different routes for the control of hand function, both in healthy primates and after recovery from lesion.

Spinal commissural connections to motoneurons controlling the primate hand and wrist.

Left-right coordination is essential for locomotor movements and is partly mediated by spinal commissural systems. Such coordination is also essential for reaching and manipulation in primates, but the role of spinal commissural systems here has not been studied. We investigated commissural connectivity to motoneurons innervating forelimb muscles using intracellular recordings in acutely anesthetized macaque monkeys. In 57 of 81 motoneurons, synaptic responses (52 of 57 excitatory) were evoked after contralateral intraspinal microstimulation in the gray matter (cISMS; 300 µA maximum current intensity). Some responses (15 of 57) occurred at latencies compatible with a monosynaptic linkage, including in motoneurons projecting to intrinsic hand muscles (9 cells). Three pieces of evidence suggest that these effects reflected the action of commissural interneurons. In two cells, preceding cISMS with stimulation of the contralateral medial brainstem descending pathways facilitated the motoneuron responses, suggesting that cISMS acted on cell bodies whose excitability was increased by descending inputs. Pairing cISMS with stimulation of the contralateral corticospinal tract yielded no evidence of response occlusion in 16 cells tested, suggesting that the effects were not merely axon reflexes generated by stimulation of corticospinal axon branches, which cross the midline. Finally, stimulation of contralateral peripheral nerves evoked responses in 28 of 52 motoneurons (7 of 9 projecting to the hand). Our results demonstrate the existence of commissural neurons with access to spinal motoneurons in primate cervical spinal cord that receive inputs from the periphery as well as descending pathways. Most importantly, commissural neurons also innervate motoneurons of intrinsic hand muscles.

Probabilistic identification of cerebellar cortical neurones across species.

Despite our fine-grain anatomical knowledge of the cerebellar cortex, electrophysiological studies of circuit information processing over the last fifty years have been hampered by the difficulty of reliably assigning signals to identified cell types. We approached this problem by assessing the spontaneous activity signatures of identified cerebellar cortical neurones. A range of statistics describing firing frequency and irregularity were then used, individually and in combination, to build Gaussian Process Classifiers (GPC) leading to a probabilistic classification of each neurone type and the computation of equi-probable decision boundaries between cell classes. Firing frequency statistics were useful for separating Purkinje cells from granular layer units, whilst firing irregularity measures proved most useful for distinguishing cells within granular layer cell classes. Considered as single statistics, we achieved classification accuracies of 72.5% and 92.7% for granular layer and molecular layer units respectively. Combining statistics to form twin-variate GPC models substantially improved classification accuracies with the combination of mean spike frequency and log-interval entropy offering classification accuracies of 92.7% and 99.2% for our molecular and granular layer models, respectively. A cross-species comparison was performed, using data drawn from anaesthetised mice and decerebrate cats, where our models offered 80% and 100% classification accuracy. We then used our models to assess non-identified data from awake monkeys and rabbits in order to highlight subsets of neurones with the greatest degree of similarity to identified cell classes. In this way, our GPC-based approach for tentatively identifying neurones from their spontaneous activity signatures, in the absence of an established ground-truth, nonetheless affords the experimenter a statistically robust means of grouping cells with properties matching known cell classes. Our approach therefore may have broad application to a variety of future cerebellar cortical investigations, particularly in awake animals where opportunities for definitive cell identification are limited.

Changes in descending motor pathway connectivity after corticospinal tract lesion in macaque monkey.

Damage to the corticospinal tract is a leading cause of motor disability, for example in stroke or spinal cord injury. Some function usually recovers, but whether plasticity of undamaged ipsilaterally descending corticospinal axons and/or brainstem pathways such as the reticulospinal tract contributes to recovery is unknown. Here, we examined the connectivity in these pathways to motor neurons after recovery from corticospinal lesions. Extensive unilateral lesions of the medullary corticospinal fibres in the pyramidal tract were made in three adult macaque monkeys. After an initial contralateral flaccid paralysis, motor function rapidly recovered, after which all animals were capable of climbing and supporting their weight by gripping the cage bars with the contralesional hand. In one animal where experimental testing was carried out, there was (as expected) no recovery of fine independent finger movements. Around 6 months post-lesion, intracellular recordings were made from 167 motor neurons innervating hand and forearm muscles. Synaptic responses evoked by stimulating the unlesioned ipsilateral pyramidal tract and the medial longitudinal fasciculus were recorded and compared with control responses in 207 motor neurons from six unlesioned animals. Input from the ipsilateral pyramidal tract was rare and weak in both lesioned and control animals, suggesting a limited role for this pathway in functional recovery. In contrast, mono- and disynaptic excitatory post-synaptic potentials elicited from the medial longitudinal fasciculus significantly increased in average size after recovery, but only in motor neurons innervating forearm flexor and intrinsic hand muscles, not in forearm extensor motor neurons. We conclude that reticulospinal systems sub-serve some of the functional recovery after corticospinal lesions. The imbalanced strengthening of connections to flexor, but not extensor, motor neurons mirrors the extensor weakness and flexor spasm which in neurological experience is a common limitation to recovery in stroke survivors.

Lack of evidence for direct corticospinal contributions to control of the ipsilateral forelimb in monkey.

Strong experimental evidence implicates the corticospinal tract in voluntary control of the contralateral forelimb. Its potential role in controlling the ipsilateral forelimb is less well understood, although anatomical projections to ipsilateral spinal circuits are identified. We investigated inputs to motoneurons innervating hand and forearm muscles from the ipsilateral corticospinal tract using multiple methods. Intracellular recordings from 62 motoneurons in three anesthetized monkeys revealed no monosynaptic and only one weak oligosynaptic EPSP after stimulation of the ipsilateral corticospinal tract. Single stimulus intracortical microstimulation of the primary motor cortex (M1) in awake animals failed to produce any responses in ipsilateral muscles. Strong stimulation (>500 µA, single stimulus) of the majority of corticospinal axons at the medullary pyramids revealed only weak suppressions in ipsilateral muscles at longer latencies than the robust facilitations seen contralaterally. Spike-triggered averaging of ipsilateral muscle activity from M1 neural discharge (184 cells) did not reveal any postspike effects consistent with monosynaptic corticomotoneuronal connections. We also examined the activity of 191 M1 neurons during ipsilateral or contralateral "reach to precision grip" movements. Many cells (67%) modulated their activity during ipsilateral limb movement trials (compared with 90% with contralateral trials), but the timing of this activity was best correlated with weak muscle activity in the contralateral nonmoving arm. We conclude that, in normal adults, any inputs to forelimb motoneurons from the ipsilateral corticospinal tract are weak and indirect and that modulation of M1 cell firing seems to be related primarily to control of the contralateral limb.

Multiple extra-synaptic spillover mechanisms regulate prolonged activity in cerebellar Golgi cell-granule cell loops.

Despite a wealth of in vitro and modelling studies it remains unclear how neuronal populations in the cerebellum interact in vivo. We address the issue of how the cerebellar input layer processes sensory information, with particular focus on the granule cells (input relays) and their counterpart inhibitory interneurones, Golgi cells. Based on the textbook view, granule cells excite Golgi cells via glutamate forming a negative feedback loop. However, Golgi cells express inhibitory mGluR2 receptors suggesting an inhibitory role for glutamate. We set out to test this glutamatergic paradox in Golgi cells. Here we show that granule cells and Golgi cells interact through extra-synaptic signalling mechanisms during sensory information processing, as well as synaptic mechanisms. We demonstrate that such interactions depend on granule cell-derived glutamate acting via inhibitory mGluR2 receptors leading causally to the suppression of Golgi cell activity for several hundreds of milliseconds. We further show that granule cell-derived inhibition of Golgi cell activity is regulated by GABA-dependent extra-synaptic Golgi cell inhibition of granule cells, identifying a regulatory loop in which glutamate and GABA may be critical regulators of Golgi cell­granule cell functional activity. Thus, granule cells may promote their own prolonged activity via paradoxical feed-forward inhibition of Golgi cells, thereby enabling information processing over long timescales.

Functional subdivision of feline spinal interneurons in reflex pathways from group Ib and II muscle afferents; an update.

A first step towards understanding the operation of a neural network is identification of the populations of neurons that contribute to it. Our aim here is to reassess the basis for subdivision of adult mammalian spinal interneurons that mediate reflex actions from tendon organs (group Ib afferents) and muscle spindle secondary endings (group II afferents) into separate populations. Re-examining the existing experimental data, we find no compelling reasons to consider intermediate zone interneurons with input from group Ib afferents to be distinct from those co-excited by group II afferents. Similar patterns of distributed input have been found in subpopulations that project ipsilaterally, contralaterally or bilaterally, and in both excitatory and inhibitory interneurons; differences in input from group I and II afferents to individual interneurons showed intra- rather than inter-population variation. Patterns of reflex actions evoked from group Ib and II afferents and task-dependent changes in these actions, e.g. during locomotion, may likewise be compatible with mediation by premotor interneurons integrating information from both group I and II afferents. Pathological changes after injuries of the central nervous system in humans and the lineage of different subclasses of embryonic interneurons may therefore be analyzed without need to consider subdivision of adult intermediate zone interneurons into subpopulations with group Ib or group II input. We propose renaming these neurons 'group I/II interneurons'.

Cerebellar Golgi cells in the rat receive convergent peripheral inputs via a lateral reticular nucleus relay.

Golgi cells are important players in the function of the cerebellar cortex, controlling the flow of incoming information from mossy fibres to the granule cells, which excite other cortical neurons. We recently showed that in anaesthetized rats most Golgi cells respond to stimulation of afferents from a very wide peripheral receptive field with a long-lasting depression of firing. These responses are mediated via a crossed ascending afferent pathway but the supraspinal part of this pathway is unknown. Here we have examined the hypothesis that the lateral reticular nucleus, a brainstem nucleus with known broad afferent convergence that projects mossy fibres to much of the cerebellum, is involved. First, we showed that single-pulse electrical microstimulation within the lateral reticular nucleus can elicit long-lasting depressions in Golgi cells, which are qualitatively similar to those evoked by peripheral afferent stimulation. Second, we showed that the amplitude of the depressions of Golgi cell firing evoked by peripheral stimulation can be reduced by pharmacological manipulation of the lateral reticular nucleus, either ipsilateral or contralateral to the stimulus site, with local injections of either the GABA(A) receptor agonist muscimol or the AMPA receptor blocker 6-cyano-7-nitroquinoxaline-2,3-dione. This evidence suggests that the lateral reticular nucleus is a relay nucleus in the brainstem for peripheral afferent information in a pathway that generates Golgi cell long-lasting depression responses.

Electrophysiological localization of eyeblink-related microzones in rabbit cerebellar cortex.

The classically conditioned eyeblink response in the rabbit is one of the best-characterized behavioral models of associative learning. It is cerebellum dependent, with many studies indicating that the hemispheral part of Larsell's cerebellar cortical lobule VI (HVI) is critical for the acquisition and performance of learned responses. However, there remain uncertainties about the distribution of the critical regions within and around HVI. In this learning, the unconditional stimulus is thought to be carried by periocular-activated climbing fibers. Here, we have used a microelectrode array to perform systematic, high-resolution, electrophysiological mapping of lobule HVI and surrounding folia in rabbits, to identify regions with periocular-evoked climbing fiber activity. Climbing fiber local field potentials and single-unit action potentials were recorded, and electrode locations were reconstructed from histological examination of brain sections. Much of the sampled cerebellar cortex, including large parts of lobule HVI, was unresponsive to periocular input. However, short-latency ipsilateral periocular-evoked climbing fiber responses were reliably found within a region in the ventral part of the medial wall of lobule HVI, extending to the base of the primary fissure. Small infusions of the AMPA/kainate receptor antagonist CNQX into this electrophysiologically defined region in awake rabbits diminished or abolished conditioned responses. The known parasagittal zonation of the cerebellum, supported by zebrin immunohistochemistry, indicates that these areas have connections consistent with an essential role in eyeblink conditioning. These small eyeblink-related areas provide cerebellar cortical targets for analysis of eyeblink conditioning at a neuronal level but need to be localized with electrophysiological identification in individual animals.

Direct and indirect connections with upper limb motoneurons from the primate reticulospinal tract.

Although the reticulospinal tract is a major descending motor pathway in mammals, its contribution to upper limb control in primates has received relatively little attention. Reticulospinal connections are widely assumed to be responsible for coordinated gross movements primarily of proximal muscles, whereas the corticospinal tract mediates fine movements, particularly of the hand. In this study, we used intracellular recording in anesthetized monkeys to examine the synaptic connections between the reticulospinal tract and antidromically identified cervical ventral horn motoneurons, focusing in particular on motoneurons projecting distally to wrist and digit muscles. We found that motoneurons receive monosynaptic and disynaptic reticulospinal inputs, including monosynaptic excitatory connections to motoneurons that innervate intrinsic hand muscles, a connection not previously known to exist. We show that excitatory reticulomotoneuronal connections are as common and as strong in hand motoneuron groups as in forearm or upper arm motoneurons. These data suggest that the primate reticulospinal system may form a parallel pathway to distal muscles, alongside the corticospinal tract. Reticulospinal neurons are therefore in a position to influence upper limb muscle activity after damage to the corticospinal system as may occur in stroke or spinal cord injury, and may be a target site for therapeutic interventions.

Characterization in vivo of bilaterally branching pontocerebellar mossy fibre to Golgi cell inputs in the rat cerebellum.

Golgi cells regulate the flow of information from mossy fibres to the cerebellar cortex, through a mix of feedback and feedforward inhibitory actions on granule cells. The aim of the current study was to examine mossy fibre input to Golgi cells, in order to assess their impact on switching Golgi cells into feedforward behaviour. In urethane-anaesthetized rats, extracellular recordings were made from Golgi cells in Crus II (n = 18). Spikes were evoked in all Golgi cells by microstimulation within the contralateral hemispheral cortex, via branches of mossy fibres that terminate in both cerebellar hemispheres. The latencies of these responses were very short, consistent with a monosynaptic mossy fibre contact [average onset latency 2.3 +/- 0.1 ms (SEM)]. The same stimuli had no measurable effect on spike responses of nearby Purkinje cells (n = 12). Systematic mapping in the contralateral cerebellar hemisphere (Crus Ib, IIa, IIb and the paramedian lobule) usually revealed one low-intensity stimulus 'hotspot' (12-35 microA) from which short-latency spikes could be evoked in an individual Golgi cell. Microinjections of red and green retrograde tracers (latex beads, approximately 50-150 nL injection volume) made at the recording site and the stimulation hotspot resulted in double-labelled neurons within the pontine nuclei. Overall, this suggests that subsets of pontine neurons supply mossy fibres that branch to both hemispheres, some of which directly target Golgi cells. Such an arrangement may provide a common feedforward inhibitory link to temporally couple activity on both sides of the cerebellum during behaviour.

Cerebellar Golgi cells in the rat receive multimodal convergent peripheral inputs via the lateral funiculus of the spinal cord.

We recently showed that the activity of cerebellar Golgi cells can be powerfully modulated by stimulation of peripheral afferents, in a pattern different to local Purkinje cells. Here we have examined the pathways underlying these responses. Graded electrical stimulation of muscle and cutaneous nerves revealed that long-lasting depressions and short-lasting excitations of Golgi cells were evoked by stimulation of cutaneous nerves at stimulus intensities that activated large mechanoreceptive afferents, and grew as additional afferents were recruited. In contrast, none of the neurones responded to stimulation of muscle nerves at intensities that activated group I afferents, although about half responded with long-lasting depressions, but not excitations, to stimuli that recruited group II and III afferents. Selective lesions of the spinal dorsal columns did not affect either of these types of response. After lesions of one lateral funiculus in the lumbar cord the responses evoked by stimulation of the hindlimb contralateral to the lesion were reduced or abolished, leaving responses evoked by ipsilateral hindlimb afferents unaltered. Since both ipsi- and contralateral afferents generate responses in Golgi cells, the convergence from the two sides must occur supraspinally. It is difficult to reconcile these properties with any of the direct spinocerebellar pathways or spinoreticulocerebellar pathways that have been described. Instead, it is likely that the responses are evoked via the multimodal 'wide dynamic range' neurones of the anterolateral system. Golgi cell activity may thus be powerfully enhanced or depressed during arousal via the anterolateral system.

Different responses of rat cerebellar Purkinje cells and Golgi cells evoked by widespread convergent sensory inputs.

While the synaptic properties of Golgi cell-mediated inhibition of granule cells are well studied, less is known of the afferent inputs to Golgi cells so their role in information processing remains unclear. We investigated the responses of cerebellar cortical Golgi cells and Purkinje cells in Crus I and II of the posterior lobe cerebellar hemisphere to activation of peripheral afferents in vivo, using anaesthetized rats. Recordings were made from 70 Golgi cells and 76 Purkinje cells. Purkinje cells were identified by the presence of climbing fibre responses. Golgi cells were identified by both spontaneous firing pattern and response properties, and identification was confirmed using juxtacellular labelling of single neurones (n = 16). Purkinje cells in Crus II showed continuous firing at relatively high rates (25-60 Hz) and stimulation of peripheral afferents rarely evoked substantial responses. The most common response was a modest, long-latency, long-lasting increase in simple spike output. By comparison, the most common response evoked in Golgi cells by the same stimuli was a long-latency, long-lasting depression of firing, found in approximately 70% of the Golgi cells tested. The onsets of Golgi cell depressions had shorter latencies than the Purkinje cell excitations. Brief, short-latency excitations and reductions in firing were also evoked in some Golgi cells, and rarely in Purkinje cells, but in most cases long-lasting depressions were the only significant change in spike firing. Golgi cell responses could be evoked using air puff or tactile stimuli and under four different anaesthetic regimens. Long-lasting responses in both neurone types could be evoked from wide receptive fields, in many cases including distal afferents from all four limbs, as well as from trigeminal afferents. These Golgi cell responses are not consistent with the conventional feedback inhibition or 'gain control' models of Golgi cell function. They suggest instead that cerebellar cortical activity can be powerfully modulated by the general level of peripheral afferent activation from much of the body. On this basis, Golgi cells may act as a context-specific gate on transmission through the mossy fibre-granule cell pathway.