THE CEREBELLUM AND COGNITION

The most recent research in the cognitive neuroscience literature shows the significant involvement of the cerebellum in a diverse range of cognitive functions. The term cognition usually refers to thought processes such as executive function, learning, memory, visual analysis and language. This research is showing that the cerebellum is actively involved in processing in all these areas of cognition. Also, researchers are finding the cerebellum is intricately linked with emotion, personality and behaviour as well as autonomic and vascular regulation.

Connectional neuro-anatomy has provided a wealth of information concerning the anatomic substrates that may support the cerebellar contribution to cognition. Functional neuro-imaging has also provided much of the impetus for the exploration of the hypotheses and concepts derived from systems neuroanatomy, theoretical modelling, and experimental observations. The demonstration of cerebellar activation by non-motor tasks, first noted incidentally and then studied as a specific entity in its own right, has essentially validated the new questions “why is the cerebellum being activated, where and under what conditions”.

The first decade of research utilising this methodology has revealed insights and posed new questions about the cerebellum. An entirely separate long-standing view of cerebellar function has been overshadowed by its role in the motor system from the earliest days of clinical case reporting. Instances of mental and intellectual dysfunction were described in the settings of cerebellar pathology.

Investigators at that time lacked the necessary clinical and pathological tools to provide a clear understanding of their patients’ lesions and psychiatric and cognitive disturbances. Consequently, their anecdotal reports have been essentially ignored. The result has been that the possibility of a causal relationship between cerebellar dysfunction and cognitive and psychiatric pathology, has either been summarily dismissed or not considered for lack of awareness of the question having been posed.

In addition to this clinical background, the substantial body of experimental evidence dating back to the early part of the nineteenth century indicates that the cerebellum is involved in a number of non-motor functions. The motor bias with respect to the study of the cerebellum has been so overwhelming that this work, albeit conducted by eminent neuro-physiologists, has been omitted from mainstream thinking about the functions of the cerebellum.

Early work has shown a close relationship between the cerebellum and the autonomic nervous system (vegetative phenomena). Stimulation of the fastigial nucleus (by Moruzzi and Magoun, 1949), and of the cortex of the cerebellar anterior lobe, produced generalised arousal of the EEG, and Snider et al, (1949), demonstrated an inhibitory influence of the cerebellum on the inhibitory part of the brain stem reticular formation. Fastigial nucleus ablation produced a state of constant hyperactivity both in the monkey and the cat, (see

Carpenter, 1959 and Sprague and Chambers, 1959), and the role of the fastigial nucleus and regulation of the sleep/awake cycle, was shown by Manzoni et al, 1968.

Later, anatomic investigations confirmed physiological data showing connections between the cerebellum and reticular hypothalamic and limbic structures. Andrew Arthur Abbie in 1934 studied the cerebro-cerebellar system, focusing on the anatomy of the cortico-pontine pathway. He observed degeneration in the cerebral peduncle and basis pontis of the human brain following large lesions involving the parietal, temporal and occipital lobes. He was intrigued by the existence of this tract connecting non-motor areas of the cerebral hemisphere with the pons. He suggested that this pathway “weaves all sensory impulses into a homologous fabric and translates the resultant in muscular response, which is accurately co-ordinated and acutely adapted to the requirements of the situation as a whole. To it, man owes the possibility of his highest powers as expressed in his work, in sport, and in art.

Robert S Dow in 1942 and 1974 determined that the dentate nucleus of the cerebellum could be divided into two components, on the basis of differential staining properties and microscopic anatomy. The lateral part of the dentate (the neodentate) is phylogenetically and more recently developed that the medial part (Dow 1942). The coincidence in evolution of the appearance of the neodentate and the expanded lateral cerebellar hemisphere with the expansion of the frontal and temporal association areas, later lead Dow (1974 and 1978) to postulate that the cerebral and cerebellar regions were anatomically interconnected, and therefore functionally relevant.

THE OUTPUT CHANNELS OF THE DENTATE.

Peripheral sensory afferents to the cerebellum have been shown from early works of Sherrington in 1906, who showed that the cerebellum receives afferents from the proprioceptive system. Dow in 1939 demonstrated that the stimulation of the sciatic and saphenous nerves in the cat, resulted in cerebellar sensory potentials and in the rat he showed (Dow and Anderson, 1942) that both proprioceptive and cutaneous stimulation also resulted in cerebellar action potentials.

Dow and Moruzzi in 1958, in their book on the cerebellum, reached the conclusion that the arrival of splanchnic or vagal volleys, and of the auditory or visual impulses to the cerebellar cortex and the possible modification of the reflex activity of vasomotor centres by cerebellifugal volleys, justify the hypothesis that a hitherto unknown control may be exerted by the cerebellum in the sensory sphere and on autonomic functions. Unfortunately no adequate tests were performed at the time to analyse this data.

Sensory, visual and auditory connections in the cerebellum were shown by Snider and Wolsey. Snider and Stowell, 1944, showed that there are topographically organised cerebellar tactile receiving areas responsive to both proprioceptive input and cutaneous stimulation. Snider and Eldred, 1948, and Snider and Stowell, 1942, also demonstrated visual and auditory projections to the cerebellar vermis and that visual projections are conveyed via the tectum (Snider, 1945).

Anatomical studies (Sunderland, 1940, Brodel and Jansen, 1946) of the feed-forward loop of the cerebro-cerebellar system and electrophysiological experiments (Henneman et al, 1948) of the feed-back limb, were also influential in shaping Snider’s conclusions. He found that there are dual projections to the cerebellum, one from end organs and one from related sensory and motor areas of the cerebral hemispheres. Snider saw the cerebellum as the great modulator of neurologic function and predicted for it a role not only in the field of neurology, but also in psychiatry.

In later work on connections linking the cerebellum with the locus ceruleus and limbic structures, hippocampus, septum, and amygdala (Snider, 1975 and 1976) supported his contention in the notion of cerebellar function needing to be revised.

Snider also noticed that not all lesions produce ataxia. He noted that one could remove considerable masses of cerebellar tissue without producing any apparent deficits. This conclusion was also apparent to Dow in 1974, who commented that it was particularly true for the lateral cerebellar cortex and the dentate nucleus. He posed the question that if lesions or cooling of the dentate nucleus alone are not productive of the classical signs of cerebellar ataxia, what methods can one employ to unravel the functions of this part of the cerebellum, which is so large in man and so selectively related to the association areas of the cerebral cortex (Dow, 1974, page 115).

In the 1970s there was an increased interest in non-motor functions of the cerebellum and a number of studies addressed different aspects. James Prescott, 1971, presented views which are similar to those of early developmental psychologists, most notably Jean Piaget.

He felt that movement is intricately bound with sensation and with intellectual and emotional growth, and Prescott reached the conclusion that the cerebellum participates in the emotional development and that it is a master integrating and regulatory system for sensory, emotional and motor processes. He asserted that maternal social deprivation of neonatal animals (the Harlow monkeys, 1971) is fundamentally a form of somatosensory input deficit. He theorises that the physiological effect of this deprivation would be a reduction in the number of cerebellar neurones, and those neurones that survived would operate under a condition of denervation supersensitivity.

He reasoned that the psycho-pathological characteristics that the animals tested manifested (rhythmic rocking, head-banging) reflects the effects of sensory de-afferentation at a critical period. He postulates that hyper-reactive cerebellar neurones generate unusual movement patterns. Compare this to similar behaviours in patients with autism.

Heath and colleagues in 1972 studied the relationship between the cerebellum and psychopathology, and used electrophysiology recordings. He demonstrated fastigial nucleus connections with the septum, as well as with the hippocampus and amygdala. Reciprocal cerebellar connections with the hypothalamus and mamillary bodies were to be convincingly shown anatomically in later studies. A cerebellar influence on human emotional experience has been shown when the dentate nucleus and the superior cerebellar peduncle were stimulated (Nashold and Slaughter, 1969).

In 1977 Heath produced amelioration of aggression in ten out of eleven patients with severe emotional dyscontrol by chronically stimulating the cerebellar vermis through subdurally implanted electrodes. Following up six and sixteen months later, ten out of the eleven patients were reported to be markedly improved. He ascribed these behavioural effects to cerebellar connections with the limbic system. Heath concluded that the cerebellum operates as an emotional pacemaker, necessary for the modulation of normal behaviour.

Berman et al, 1978, concluded that the vermis and archicerebellum are concerned with aggression. Cooper and colleagues in 1974 and 1978, demonstrated that cerebellar cortical stimulation achieved seizure control in his patients, but also had the unexpected side-effect of improving aggression, anxiety and depression.

A number of experimental observations in the 1980s in the many disciplines within the neurosciences helped anchor the role of the cerebellum in non-motor and cognitive processes. Classically conditioned learning was shown to be dependent on the cerebellum by Thompson in 1988 with his rabbit nictitating response. Leaton and Supple, 1986, showed the cerebellar involvement in the acoustic startle response of the rat and visual spatial navigational skills were impaired in the mutant mouse model by Lalonde and Botez, 1986.

Anatomic substrates for the cerebellum’s modulation of cognitive processing began to be demonstrated by the findings of organised projections from associative and paralimbic cerebral areas, to the feed-forward limb of the cerebro-cerebellar system (Schmahmann and Pandya, 1987, 1989). Dow, 1974, demonstrated a relationship between the neodentate nucleus and the prefrontal cortex during his collaboration with Leiner et al in 1986 and 198. A neuropathologic study, Bauman and Kenper, 1985, and neuroimaging (Courchesne et al, 1988) of patients with early infantile autism, revealed abnormalities in the cerebellum, and these anatomic correlations and their clinical relevance remains the subject of ongoing study.

The cerebro-cerebellar system fits in with electro-cerebellar function and cognition. If there is a cerebellar contribution to non-motor function and particularly to cognitive abilities and effective states, then there must be a corresponding anatomic substrate that supports this: this is the cerebro-cerebellar circuit.

This circuit consists of a feed-forward or afferent limb and feed-back, or efferent limb. The feed-forward limb to the cerebellum is composed of cortico-pontine and ponto-cerebellar mossy fibre projections.

CEREBROCEREBELLAR CIRCUIT.

The feed-back loop is the cerebello-thalamic and thalamo-cortical pathways linking the cerebellum to the cortex. Schmahmann in 1991, developed a conceptual approach stating that the cerebellum modifies behaviourally relevant information that it has received from the cerebral cortex via the cortico-pontine pathways. It then re-distributes this cerebellar-processed information back to the cerebral hemispheres, and therefore both limbs are essential for discussion of cognitive and non-motor processing.

There is a second feed-forward system linking the cerebral cortex with the red nucleus, from where the central tegmental tract leads to the inferior olivary nucleus, and then through the climbing fibre system to the cerebellar cortex. This second afferent arc has more restricted relevance to the discussion of the relationship between the cerebellum and cognition.

The feed-forward limb of the cerebro-cerebellar system consists of the cortico-pontine projections. These projections come from several areas of the cortex into the basilar pons and are the obligatory first stage in the feed-forward limb of the cerebro-cerebellar loop. These cortico-pontine pathways originate in neurones in layers Vb of the cerebral cortex. The axons enter the internal capsule and descend into the cerebral peduncle and terminate around neurones that occupy the ventral half of the pons.

THE FEEDBACK LIMB OF THE CEREBROCEREBELLAR CIRCUIT.

The basilar pons appears to be parcelated into different nuclear groups, (see Schmahmann and Pandya, 1989 and 199)1, and appears to indicate different areas of cerebral projection. These projections appear to inter-digitate with each other, but do not overlap. Motor, pre-motor and supplementary motor regions, as well as primary somatosensory cortices have been shown to send their efferents to the cerebellum via this route. However, the origins of

the cortico-pontine pathway are not limited to the sensory motor cortices. The cerebral cortex areas of interest in regard to cognitive ability are areas of the association cortex of the parietal, temporal and frontal lobes, which are responsible for highly complex cognitive operations and when lesioned in humans, result in clinical syndromes which are now part of classical neurological teaching. Also, the paralimbic areas of the para-hippocampal gyrus and cingulate cortices, are concerned with motivation and drive, and are thought to play a role in emotionally relevant memory (Nadel, 1991).

PARIETO-PONTINE CONNECTIONS

These connections are critical for directed attention, visio-spatial analysis and vigilance in the contra-lateral hemi-space and lesions are associated with disturbances of complex visio-spatial integration, trimodal neglect of the contra-lateral body and extra-personal space.

ALIEN HAND SYNDROME, IMPAIRED LANGUAGE, APRAXIA AND AGNOSIA

The superior parietal lobule (SPL) and the inferior parietal lobule (IPL) are thought to be involved in the sequential processing of somatotopically organised information received from adjacent primary somatosensory cortices and this includes somatosensory as well as visual and vestibular information.

MULTIPLE JOINT POSITION, SENSE, TOUCH AND PROPRIOCEPTION

These areas have connections with the prefrontal cortex and the cingular gyrus as well as paralimbic cortices and other multi-modal zones in the temporal and frontal lobes. Recent studies (Glickstein et al, 1985, May and Anderson, 1986, Schmahmann and Pandya, 1989) have shown that there are consistent pontine projections from these regions. These projections are directed most heavily towards the peri-peduncular and lateral nuclei.

TEMPORO-PONTINE CONNECTIONS

The role of the temporal lobe with respect to language, memory and complex behaviour has been well established and confusional states, highly structured visual hallucinations and the Cluver-Bucy syndrome, consequent upon lesions in this area, are also recognised. The cortex in the upper bank of the superior temporal sulcus (STS) has been shown to be concerned with multiple sensory modalities; vision, somatic sensation and audition. It has connections with association areas of the frontal and parietal cortices, as well as with limbic related structures at the medial and inferior frontal convexity and para-hippocampal and cingulate gyrus.

The superior temporal gyrus appears to be an association area confined to the auditory realm and the deep area of the STS is the important association area for the somatosensory modality. The extreme dorso-lateral and lateral pontine nuclei are the major recipients of

efferents from each of the sub-divisions of the upper bank of the STS. The inferior temporal region and the lower bank of the STS, which are strongly interconnected, contain neurones that are functionally unimodal within the visual system. They subserve mainly central vision and seem to be involved in object recognition. They appear to contribute no projections to the basis pontes and it appears that visual projection to the pons is derived exclusively from areas devoted to the peripheral visual field rather than from areas subserving central or focal representation.

PARA-STRIATE, OCCIPITO-TEMPORAL AND PARA-HIPPOCAMPAL PROJECTIONS TO THE PONS

The dichotomy in the pontine connectivity between the visual motion “where” versus visual feature “what” system is also observed in the para-striate and occipito-temporal pontine system. The pons receives the dorsal visual stream responsible for spatial features of objects or events in the periphery of the visual field and these are distributed in the lateral peri-peduncular and dorso-lateral pontine nuclei. The ventral stream, which deals with form, colour and orientation and with the stimuli occurring in the central part of the visual field, does not project to the pons.

It also appears that only the posterior para-hippocampal regions project to the pons and these are concerned again with the peripheral rather than central visual field and are concerned with the spatial aspects of memory. These afferents appear to facilitate a cerebellar contribution to visual spatial memory, particularly when invested with motivational valence.

PRE-FRONTO-PONTINE CONNECTIONS

The prefrontal cortex has repeatedly been shown in both humans and non-human primates to be an essential component in the normal integration of higher order behaviour. This includes such functions as planning, foresight, judgement, attention, language and working memory. Different functional attributes have been ascribed to different regions, but also lateral and medial convexities are important for kinesthetic, motivational and spatially related functions, including spatial memory, whereas inferior prefrontal and orbital areas are more related to autonomic and emotional response; inhibition, stimulus significance and object recognition and memory. Schmahmann and Pandya in 1995 and 1997, demonstrated that connectional heterogeneity of the prefrontal cortex also exists in the cortico-pontine pathway.

The dorsal lateral convexity and the medial prefrontal cortex provide the majority of the pontine efferents. The projections are most prominent and occupy the rostro-caudal extent of the pons when derived from dorsal area 46 (9/46d), areas 8a,d and 9 at the dorso-lateral convexity and area 10 at the dorso-lateral and medial convexities.

The topographic organisation within this general framework of projection is discernible. More medial prefrontal areas send projections to the most medial pontine regions, whereas

pontine terminations tend to shift away from the midline following lateral prefrontal projections. Furthermore, each cortical area appears to have a unique complement of pontine nuclei with which it is connected.

The fronto-pontine connections are concerned with attention as well as with conjugate eye movements (area 8a), the spatial attributes of memory and working memory (area 46), planning, foresight and judgement (area 10), motivational behaviour and decision-making capabilities (area 9 and 32), and from areas considered to be homogenous to the language area in a human (area 45b).

Not all regions of the prefrontal cortex project to the pons. These areas resemble the infero-temporal region and the ventral bank of the superior temporal sulcus with which they are interconnected. They are concerned with object memory, feature discrimination and certain aspects of motivation. The dichotomy in the spatial (where) versus object (what) which are apparent in the cortico-pontine projections from other associated areas, appear to be conserved to the prefrontal areas as well.

PARA-LIMBIC AND AUTONOMIC CONNECTIONS WITH THE PONS

A direct and reciprocal hypothalamo-cerebellar projection has been identified in the monkey. The ansiform and paramedian lobules and the paraflocculus are connected with the lateral and posterior hypothalamic areas. The anterior lobe is connected with these as well as the ventro-medial and dorso-medial and dorsal hypothalamic nuclei and deep cerebellar nuclei project to the contra-lateral posterior and lateral hypothalamic nuclei.

HYPOTHALAMIC PROJECTIONS TO THE CEREBELLUM.

Also the mamillary and super-mamillary nuclei project to the cerebellar ansiform and paramedian lobules, paraflocculus and anterior lobe. Schmahmann and Pandya in 1993, demonstrated projections to the pons from the posterior para-hippocampal regions, which have been implicated in the spatial aspects of memory.

It is therefore apparent that the ponto-cerebellar system indeed receives a sizeable input from limbic related cortices and these findings may help explain the autonomic phenomena produced in animals by cerebellar stimulation, and also provide a plausible anatomic substrate for a cerebellar role in modulation of affect. The course of these fibre pathways to the pons appear to consist of segregated and partially overlapping pathways, which are to some extent distinguishable anatomically at each stage of their trajectory from their origin to destination.

There is only limited information about the exact connections between the ponto-cerebellar projections. Nevertheless, it has been established, for example, from physiological studies, that the parietal and prefrontal cortices are functionally related to the neocerebellar hemispheres and auditory and visual inputs are received in vermal lobules 6 and 7.

Anatomic and physiological studies in the monkey indicate that the dorsal paraflocculus, the uvula and the vermal visual area (vermal lobule 7) receive information from visually responsive neurones in the dorso-lateral pontine region and the nucleus reticularis tegmenta pontes.

In 1979 Brodel performed horseradish peroxidase (HRP) retrograde labelling on the ponto-cerebellar projections in the monkey. He found that the anterior lobe (lobe 5) receives input from medial parts of the caudal pons and the vermal visual area (lobules 7 and 8a), from two cell groups located in the dorso-medial and dorso-lateral pons. Vermal lobules (8b) receive input from the intrapeduncular nucleus, crus 1 from medial parts of the rostral pons and crus 2 from the medial, ventral and lateral pons.

The hemispheres have relatively greater pontine input than the rostral vermis. Brodel concluded that the anterior lobe and lobulus simplex (lobes 1 to 6) receive afferents from the motor and pre-motor cortices and, to a small extent, from the parietal lobe. The pre-motor and prefrontal cerebral regions are linked with crus 1 (lobule 7 and 8). The motor cortex is linked with crus 2 and somatosensory and parietal association areas are linked with the paramedian lobule.

Brodel concluded that there was a high degree of order with each cerebellar sub-division receiving input, at least partly, from its own pontine territory. One small part of the cerebellum receiving input from several discrete pontine cell groups situated far apart and Brodel concluded that information from one small part of the cerebral cortex is distributed to numerous discrete sites in the cerebellar cortex, where it is combined with other specific kinds of information. However, overall, detailed understanding of the ponto-cerebellar system is still not available.

Much remains to be elucidated regarding the details of the pontine afferents to defined regions of the cerebellum and with respect to the cerebral and cerebellar connections of individual basilar pontine regions. There is not yet any information regarding the transfer of associated information from the pons to the cerebellum. Higher order information is distributed in complex but specific patterns throughout the basilar pons, but the manner in which this information is conveyed to the cerebellum and the corresponding organisation within the cerebellum have not yet been studied. Also, the fractured somatotopy that has been discerned in the sensory afferents to the cerebellum may apply also to the associative system.

THE FEED-BACK LIMB OF THE CEREBRO-CEREBELLAR SYSTEM

This consists of cerebellar cortico-nuclear projections, efferents from deep cerebellar nuclei en passant through the red nucleus to the thalamus and the thalamo-cortical relay. Despite the homogenous appearance of the cerebellum, there is increasing evidence that there are neurotransmitter/modulator/peptide differences in neuronal sub-types of cerebellar cortex (Oertel, 1993).

There is a mediolateral zonal pattern of organisation of the cortex that correlates with connectional specificity in the olivary projections to the cerebellum, and this suggests that the otherwise homogenous appearing cortex can be sub-divided by methods other than gross anatomic descriptions and topographically organised connectional relationships. The cortico-nuclear projection consists of axons of the cerebellar Purkinje cells being the only neurone responsible for efferents from the cerebellar cortex that traverse the cerebellar white matter and terminate in the deep cerebellar nuclei.

The topographic arrangement of the cortico-nuclear projection is such that the midline cortex projects to the medial nuclear regions (fastigial), lateral hemispheres project to the dentate and the intervening cortex corresponds with the interposed nuclei in a predictable mediolateral pattern. The flocculonodular lobe additionally has direct connections with the vestibular nuclei and the anterior interpositus with the red nucleus. It is thought that thalamic projections from the cerebellum, which traditionally were thought to come purely from the dentate nucleus, are now thought to be assisted by efferents from the fastigial and interpositus nuclei.

More detailed understanding is needed regarding the precise topographical relationships between each cerebellar nucleus and its corresponding complement of thalamic terminations. Certain principles of organisation of the cerebello-thalamic projections have been defined however. There appear to be differential anterior versus posterior dentate nucleus projections to the thalamus and each cerebellar nuclear region projects a few (between three and seven) rostro-caudally orientated rod-like aggregates situated within a dorso-ventral curved lamella in the thalamus.

Classic cerebellar recipient motor thalamic nuclei are not alone in receiving input from the cerebellum. Non-motor nuclei have a considerable cerebellar input as well. These include the intralaminar nuclei, particularly centralis lateralis (CL) as well as the paracentralis (PCN) and the centromedian-parafascicular (CM-PF) complex and the medial dorsal nucleus. The CL nucleus appears to project in a widespread fashion as many other intralaminar nuclei do, to areas of the posterior parietal cortex (PPC), the upper bank of the superior temporal sulcus (STS), the prefrontal cortex (PFC), the cingular gyrus and the primary motor cortex. The PCN nucleus projects to the parahippocampal gyrus. The medial dorsal thalamic nucleus, which is the major site of thalamic connections with the frontal lobe, also receives cerebellar input.

Also traditionally motor thalamic nuclei have projections to regions of the cerebral cortex outside the primary and the supplementary motor areas, including the prefrontal peri-arcuate cortex, see later notes regarding area 46 of the prefrontal lobe by Middleton and Strick in 1994.

DENTATE NUCLEUS PROJECTIONS

Further research in the use of direct trans-neuronal techniques remain to show how much of the cerebellar input to the thalamus is conveyed to association cortices. Nevertheless, it would appear from available anatomic evidence that the cerebellar recipient ‘motor thalamic nuclei’ project not only to the motor cortices, but also to the associative areas in the posterior, parietal, superior, temporal and prefrontal cortices. Furthermore, the intralaminar nuclei which are themselves the recipient of cerebellar efferents and are non-motor, project widely throughout the cerebral cortex, including the motor, associative and para-limbic regions.

A CLIMBING FIBRE SYSTEM AND COGNITION

A central feature of the Marr, 1969 and Albus, 1971, theory of motor learning, is the interaction between mossy fibre and climbing fibre systems. It has been suggested that learning is an important mechanism whereby the cerebellum also modulates non-motor behaviour. Mossy fibres to the cerebellum arise largely from neurones in the basilar pons. The inferior olive is the sole source of the climbing fibres input to the cerebellum. The cerebral afferents of the pontine (mossy fibre) and olivary (climbing fibre systems) are markedly different.

The pontine system input is derived in a large part from the cerebral hemispheres including the association areas. In the non-human primate the inferior olive receives much of its descending input from the parvicellular red nucleus. Afferents of the parvicellular red nucleus are derived most heavily from motor, pre-motor and supplementary motor cortices, and to some extent from the post-central gyrus and area 5 in the superior parietal lobule. They are not derived to any convincing degree (at least in studies to the present date) from the associative or para-limbic cortices.

The zona incerta, which projects to the inferior olive has been reported to receive projections from prefrontal cortices, suggesting there maybe some indirect prefrontal input to the olivary system of climbing fibres. Schmahmann and Pandya have investigated this

and performed a review of previously performed anterograde trace experiments in the monkey, and this reveals there are prominent and topographically organised projections from the pre-central motor cortex to the parvicellular and magnocellular divisions of the red nucleus.

Additionally, there is substantial input from the supplementary motor area to the parvicellular division. However, no projections to the red nucleus were seen to arise from associative or para-limbic cerebral cortices. Significant projections to the zona incerta were observed from the cingulate cortex, as well as from the posterior parietal, prefrontal and para-striate association areas.

Additionally, prefrontal cortex projections to the zona incerta arose from areas 9/46d, as well as from area 9 and medial convexity. Therefore, the possibility of interaction between mossy fibre and climbing fibre systems in learning non-motor tasks is therefore maintained by virtue of the associative projections to the zona incerta, which in turn projects to the inferior olivary nucleus.

CEREBELLAR OUTPUT CHANNELS

Middleton and Strick have been investigating the cerebellar output channels using retrograde trans-neuronal transport of herpes simplex virus type 1 (HSV1). Two strains used interestingly show anterograde and retrograde transport. Strain H129 shows an anterograde trans-neuronal transport and McIntyre-B strain shows retrograde trans-neuronal transport. Their results suggest that the cerebellar output projects via the thalamus to multiple cortical areas, including pre-motor and prefrontal cortex, as well as the primary motor cortex.

In addition, the projections to different cortical areas appear to originate from distinct regions of the deep cerebellar nuclei and their observations lead them to propose that this cerebellar output is composed of a number of separate output channels. In the past a number of technical limitations have made it difficult to define cerebello-thalamo-cortical circuits. For example, most studies that examined the pattern of cerebellar terminations in the thalamus did not determine the cortical targets of these thalamic nuclei.

In addition, the lack of standard criteria for defining thalamic borders and confusing thalamic nomenclature have made comparison of the results from different studies difficult. Middleton and Strick developed a tracing technique with trans-neuronal transport of herpes simplex virus type 1. Their results indicate that cerebellar output targets not only the primary motor cortex, but also several areas of pre-motor, ocular motor and prefrontal cortices. They propose that the output from the cerebellum and specifically that from the dentate nucleus contains multiple output channels, each projecting to a distinct cortical area, see Strick, 1993, Middleton and Strick, 1994 and 1996.

Trans-neuronal transport of HSV1 provides a novel method for labelling a chain of synaptically linked neurones and the technique is capable of identifying circuits of at least three neurones in length.

They tested the virus originally by injecting the arm area (M1 of the motor cortex) in Cebus monkeys. After three days the virus had migrated to second order neurones in regions of the pontine nuclei known to receive input from the arm. Five days after the injection multiple patches of third order neurones were found in the cerebellar cortex. These patches were located in the granular layer and contain two types of labelled neurone granule and Golgi cells. Both these cell types are known to be contacted via mossy fibre afferents that project to the cerebellar cortex from the pontine nuclei. The majority of the labelled patches were located in the vermal and hemispheric lobules of 5 and 6, in and adjacent to the primary fissure.

Separate labelled patches were found posteriorly in the paramedian lobule 8a and laterally in lobule 7b. Also some neurones were found in portions of the dentate and interpositus nuclei. These areas corresponded to sites where evoked potentials have been recorded after stimulation of the arm area of the primary motor cortex using McIntyre-B strain retrograde analysis of the cerebello-thalamo-cortical tracts. Using this type of tracking, Middleton and Strick have observed cerebellar projections to the primary motor cortex, ventral pre-motor area (area 6), the frontal eye field, FEF (area 8) and two regions in the prefrontal cortex (area 9 and 46).

CEREBELLAR OUTPUT TO PREFRONTAL CORTEX

Middleton and Strick have looked at Brodmann areas BA9 and BA 46. These areas have been reported to be involved in working memory and in the guidance of behaviour based on transiently stored information rather than immediate external cues. Both have been shown to project to regions of the pontine nuclei and to receive input from sub-divisions of the ventro-lateral thalamus. Viral injections into prefrontal cortex labelled many neurones in the dentate nucleus. These neurones were confined to the most ventral portions of the dentate and were concentrated rostro-caudally in the middle third of the nucleus. Therefore, these regions of the dentate clearly differ from the more dorsal regions that were labelled by virus injections into the motor cortex and also the more caudal regions of the dentate labelled by injections in the frontal eye fields.

Two conclusions arise from these results. Firstly the output of the cerebellum can influence skeletal motor, ocular motor and prefrontal regions of the cerebral cortex. Secondly, each of these different cortical regions receive input from a different region of the dentate, and therefore the dentate appears to contain a number of distinct output channels, which project via the thalamus to specific areas of the cerebral cortex. Middleton and Strick also performed physiological studies examining the function of the dentate nucleus during a learned behaviour mechanism in monkeys. They were able to analyse activity in the motor and supplementary motor and prefrontal areas separately by analysing a sequence of learned tasks. They were able to discriminate between purely motor function, supplementary motor function involved in more complex higher order aspects of motor behaviour, such as the generation of sequential movements based on external cues, and

also output channels that influence the prefrontal cortex involved in cognitive aspects of behaviour, such as working memory. These systems seem to have separate output channels in different areas of the dentate nucleus.

All these studies were performed on human subjects using magnetic resonance imaging (MRI). There were two tasks; one using a visual guided motor task using the dominant hand to move pegs from one hole to another. Another task required cognitive involvement using rules for moving pegs around on the board. In the visually guided task there was minimal activation only on the ipsilateral dentate nucleus. In the cognitively determined test, both dentate nuclei were activated bilaterally and the degree of activation was three to four times larger than during the visually guided task. Additionally the ventral portions of the dentate activated by the cognitive task appeared to differ in the location from the portions of this nucleus activated during the visually guided task.

The conclusion to this experiment is that the cognitive demands associated with attempts to solve the cognitive task lead to dentate activation. Secondly, the ventral regions of the dentate involved in cognitive processing are distinct from the dentate regions involved in the control of eye and limb movements and are potentially within an output channel that innervates prefrontal cortex.