Cerebellum

A brain organ at the rear of the brain overlying the brainstem, composed of a cortex and a core of white matter, which contains deep nuclei, interconnected with all major subdivisions of the central nervous system, and involved in sensorimotor and cognitive function.

The conspicuous location and distinct shape of the cerebellum (Latin for little brain) had attracted the attention of neuroanatomists long ago; at the time it was even considered as the seat of the soul, although this was soon rejected: '... this marvelous instinct, which is developed by education into mind, and which always has its seat in the brain . and never in the cerebellum; for I have often seen the cerebellum injured... when the soul has not ceased to fulfill its functions' (La Mettrie 1748). Others thought that the spirits which put the body in motion flow from the cerebrum for voluntary movements, but from the cerebellum for unconscious movements (Hoffman 1695, cited in Brazier 1984). This latter proposal (spirits excluded) was not too far from certain more recent notions of cerebellar function. But the picture is changing. And so is the view of the role of the cerebellum in learning.

The cerebellum in mammals constitutes only about 10% of the total volume of the brain, yet contains more than half of all the neurons (Ghez 1991). It is composed of a thin cortex and a core of white matter containing pairs of deep nuclei (the fastigial, anterior and posterior interposed, and dentate nuclei; Figure 7). The cerebellar input and output pathways course through the cerebellar peduncles that connect the cerebellum to the brainstem. The cerebellar input is carried by two fibre systems, mossy fibres and climbing fibres, which reach the cerebellar cortex and send collaterals to the deep nuclei. The mossy fibres originate in the brain-stem. The climbing fibres originate in the inferior olive in the medulla, itself receiving input from the spinal cord and the cerebral cortex. The cerebellar cortex is neatly organized in three layers, containing altogether only a few types of neurons (Figure 8): Purkinje cells (large GABAergic neurons), granule cells (small *glutamatergic neurons), Golgi cells (GABAergic/ glycinergic), and stellate/basket cells (GABAergic) (Voogd and Glickstein 1998).1 Each Purkinje cell receives converging input from a large number of mossy fibres, via parallel fibres sent by the granule cells. In contrast, each Purkinje cell receives input from only one climbing fibre. The Purkinje cells, the only cerebral cortex efferent system, project inhibitory connections to the deep nuclei, which relay the cerebellar output further.

Over the years, the relative simplicity and quasi-crystalline microstructure of the cerebellar cortex has enticed a wide variety of investigators, ranging from neuroanatomists to cell biologists to theoreticians, to propose *models for cerebellar function (e.g. Braiten-berg 1967; Marr 1969; Albus 1971; Ito 1972, 1984; Eccles 1973; Raymond et al. 1996; Braitenberg et al. 1997). Two points concerning these models are particularly noteworthy. First, the models consider the cerebellum to be either an orchestrator of motor function,2 a motor learning machine, or a clock. It is now evident

Fig. 7 A simplified macroscopic view of the exposed cerebellum. In this dorsal view, the cerebral cortex is depicted as transparent to show the deep nuclei, and the right hemisphere is cut out to show the underlying cerebellar peduncles. (Adapted from Niewwenhuys et al. 1988.)

Fig. 8 A simplified microscopic view of the cerebellar cortex. Only five types of neurons (Purkinje, granule, Golgi, stellate, and basket) are organized into the three layers of the cerebellar cortex.The sole output, which is inhibitory, is provided by the Purkinje cells. Input reaches the cerebellum via two excitatory fibre systems, the mossy and the climbing fibres. Each Purkinje cell receives converging input from a large number of mossy fibres via many parallel fibres that are sent by the granule cells. In contrast, each Purkinje cell receives input from only one climbing fibre. Inset: A scheme depicting the convergence of the two major cerebellar inputs on the Purkinje cell. The influential Marr-Albus *model (Marr 1969; Albus 1971) proposed that the climbing fibre instructs the Purkinje cell to respond specifically to the concurrent pattern of parallel fibre activity; the modified synapse is encircled. (Adapted from Ghez 1991.)

Golgi cell

Fig. 8 A simplified microscopic view of the cerebellar cortex. Only five types of neurons (Purkinje, granule, Golgi, stellate, and basket) are organized into the three layers of the cerebellar cortex.The sole output, which is inhibitory, is provided by the Purkinje cells. Input reaches the cerebellum via two excitatory fibre systems, the mossy and the climbing fibres. Each Purkinje cell receives converging input from a large number of mossy fibres via many parallel fibres that are sent by the granule cells. In contrast, each Purkinje cell receives input from only one climbing fibre. Inset: A scheme depicting the convergence of the two major cerebellar inputs on the Purkinje cell. The influential Marr-Albus *model (Marr 1969; Albus 1971) proposed that the climbing fibre instructs the Purkinje cell to respond specifically to the concurrent pattern of parallel fibre activity; the modified synapse is encircled. (Adapted from Ghez 1991.)

that these possibilities are not mutually exclusive. Second, a highly influential model has proposed a locus of *plasticity in the cerebellar cortex as the key to cere-bellar learning (Marr 1969; Albus 1971). This locus is the *synapse from the parallel fibres to the Purkinje cell (Figure 8). The basic idea is that information arriving via the climbing fibre conditions the Purkinje cell to respond specifically to the concurrent pattern of parallel fibre activity. Certain lines of experimental evidence support this prediction.

Much of the data on the role of cerebellar circuits in plasticity and learning stems from two *systems. One is the adaptation of the vestibulo-ocular reflex (VOR; du Lac et al. 1995; Ito 1998); the other is *classical conditioning of the eyeblink reflex (Yeo and Hesslow 1998; Steinmetz 2000; Thompson et al. 2000).3 The basic motor pathways for both reflexes are in the brainstem, but their use-dependent modification relies on other brain structures, particularly the cerebellum. The VOR evokes eye movements in the direction opposite to head movement. It stabilizes vision, by keeping images from slipping across the retina. The reflex is capable of remarkable adaptation, e.g. after wearing reversing prisms, eye movements go with, instead of against, head movement. The visual information and the vestibular information involved in this adaptation converge both in the deep cerebral nuclei and in the cerebellar cortex. Although significant pieces of the puzzle are still missing, it is now believed by most authors that both these sites are involved in the adaptation. Furthermore, it is proposed that the convergence on the Purkinje cell of the visual information, mediated by the climbing fibres, with the vestibular information, mediated by the mossy-fibre—parallel fibre, induces long-term depression (LTD) in the parallel fibre—Purkinje cell synapse (Lisberger 1998; Ito 2001).4 This LTD is considered to contribute to the behavioural change of the VOR.

The other system that has contributed tremendously to our knowledge about the role of the cerebellum in behavioural plasticity is classical conditioning of the eyeblink reflex, in which the *subject blinks in response to a noxious stimulus applied o the eye area. This is one of the most studied cases of simple learning in mammals. It has been studied in multiple protocols and species. A popular preparation involves the response of the external eyelid or of the nictitating membrane in the rabbit.5 Commonly in this preparation the conditioned stimulus is a tone, and the unconditioned stimulus is an air puff to the eye (Figure 9). With training, the rabbit learns to blink in response to the tone alone. It can be shown, using ablations, that elemental delay conditioning of eyeblink is dependent on the cerebellum but independent of the *cerebral cortex and *hippocampus.

Studies using cellular physiology, anatomical, pharmacological, and *neurogenetic lesions, and electrical stimulation of discrete brain sites (*criterion, *method), have identified the cerebellar cortex and the interpositus nuclei in the cerebellum as sites of *associ-ation and plasticity in this type of learning. Additional learning-related plasticity changes can, however, be detected in other sites in the brain, particularly if the conditioning protocol becomes more complex than elemental conditioning. Furthermore, when a trace conditioning protocol is used instead of delay conditioning (Figure 14, p. 46), the hippocampus becomes obligatory for conditioning (e.g. Moyer et al. 1990). There is also evidence that trace conditioning of the eyeblink reflex, as opposed to delay conditioning, requires *conscious awareness (Clark and Squire 1998), which should be expected to engage cerebral cortex as well.

Though controlling different behaviours, and involving different brainstem nuclei, cerebellar cortex

Mossy fibres n

.uditory nuclei

Cerebellar cortex k—fr-1

Mossy fibres

.uditory nuclei

nuclei

Other targets "

Interpositus nucleus

Other targets "

Climbing fibres

Inferior olive

Red nucleus a.

L_r- : 1 Corneal r^cmufh'rpuff

Cranial motor nucleus

Reticular formation

Eyeblink UR&CR

Fig. 9 A flowchart diagram of the circuits that subserve classical conditioning of eyeblink in the rabbit. In a typical protocol, tone is the conditioned stimulus and periorbital air puff the unconditioned stimulus. The unconditioned stimulus elicits closure of the eyelid as well as extension across the cornea of the nictitating membrane (the internal eyelid). The cerebellar cortex and the interpositus nuclei are sites of 'association and 'plasticity in this type of learning. When trace instead of delay conditioning, or complex protocols of 'classical conditioning are used, additional brain organs, such as the *hippocam-pus, become obligatory as well ('criterion). -> , excitatory *synapse; -|, inhibitory synapse. (Adapted from Thompson and Kim 1996.)

areas and deep cerebral nuclei, the use-dependent modification of the VOR and of the eyeblink share a lot in common (e.g. Raymond et al. 1996). In both cases, as noted above, the basic reflex does not require the cerebellum, but modification of the reflex does. We hence encounter tiers of neuronal circuits, the core of which subserves the most elementary form of the behaviour while the others add adaptability. This is particularly evident in eyeblink conditioning, where the more complex forms of conditioning engage additional brain areas. In both cases, the conditioned input—vestibular in the VOR, auditory in the standard protocol of eye-blink conditioning—projects in parallel to the cerebel-lar cortex and to the deep nuclei. In both cases, the teaching input—visual in the VOR, somatosensory in eyeblink conditioning—has the opportunity to converge with the conditioned input in both the cerebellar cortex and the deep nuclei (*coincidence detector). In both cases, the convergence in the cerebellar cortex could result in modification of the Purkinje response to the conditioned input. The modified Purkinje output, via the cerebellar nuclei, may guide the motor circuits to adjust the reflex to the new conditions. Finally, in both cases, the relative contribution of the cerebellar nuclei and the cerebellar cortex, respectively, to learning, and the site(s) of the lasting memory trace, are still debated.

Recent evidence, including "functional neuroimag-ing data, unveils cerebellar involvement in "declarative memory, "working memory, and language (Leiner et al. 1993; Desmond and Fiez 1998; Thach 1998; Wiggs et al. 1999). The role of the cerebellum in these higher brain functions is possibly based on the same computational abilities, such as the spatiotemporal orchestration of strings of events, which contribute to the adaptive control of simple motor reflexes. Whatever the computations are, it becomes clear that depiction of the cerebellum merely as a motor centre is outdated.

Selected associations: Acquisition, Classical conditioning, Engram, Performance, Skill

'GABAergic neurons are so called because they release the inhibiting Neurotransmitter y-aminobutyric acid (GABA). Glycinergic cells release glycine (see *glutamate).

2Proposals that the cerebellum is involved in the control of posture and movement surfaced in the literature already in the seventeenth century (Brazier 1984). These proposals were augmented and substantiated in the nineteenth century, and validated after the First World War by reports on the behavioural consequences of cerebellar injuries (e.g. Holmes 1930). Though most current models consider the cerebellum to control movement, there is also the view that it is the inferior olive in the medulla rather than the intrinsic cerebellar circuits that is the pacemaker and spatial organizer of movement (Welsh and Llinas 1997).

3For examples of additional experimental systems that are used to investigate cerebellar function and plasticity, see Kitazawa et al. (1998),Thach (1998).

4On LTD, see *long-term potentiation, *metaplasticity. 5The nictitating membrane is an internal eyelid, a curved plate of cartilage covered with glandular epithelium, which is drawn from the inner canthus laterally across the cornea when a noxious stimulus is applied to the eye. The movement of both external and internal eyelids is controlled by nuclei in the brainstem.

0 0

Post a comment