The Cerebellum: Brain for an Implicit Self (41 page)

5.
Internal models.
Another current issue is determining how cerebellar forward and inverse models share their roles in the “real” cerebellum. There are only a small number of microcomplexes for which model properties have been defined. On the other hand, consideration of anatomical connections points to the possibility that the primary motor cortex (area 4) and the prefrontal cortex (areas 9 and 46) operate with a forward model, whereas the frontal eye field (area 8) seems to operate with an inverse model. Close comparison of these two systems should prove to be fruitful.

6.
Evolution in cerebellar circuits and their functional roles.
The large cerebellar mass contains numerous microcomplexes, which are in general designed uniformly throughout the cerebellum. However, there are also certain regional differences in the fundamental wiring of microcomplexes. These seem to be closely related to evolution. That is, the V/C-type of microcomplexes prevails in the evolutionarily old part of the cerebellum (
Chapter 12
, “
Adaptive Control System Models
,”
Section 8
). In contrast, the evolutionarily newer lateral zones of the cerebellum are characterized by the cerebrocerebellar communication loop, which provides an anatomical basis for the internal forward model (
Figure 8A
). The D
₁
-zone, which is prominent in monkeys, is linked to the premotor cortex via the dorsal part of the dentate nucleus and the thalamus (
Figure 49
). These microzones appear to provide internal models of the body schema for the control of motor actions (
Chapter 16
, “
Motor Actions and Tool Use
”). The D
₂
-zone, which is developed particularly well in humans, is linked to the prefrontal cortex via the ventral part of the dentate nucleus and it likely forms internal models of a mental model or Piaget’s schema in the control of cognitive functions (
Chapter 17
, “
Cognitive Functions
”). Lying between the old and new areas of the cerebellum, the intermediate hemisphere (C
₁
–C
₃
-zones) contains two types of microcomplex; an adaptive controller for reflexes and an internal model for voluntary movement (
Figures 33
and
46
). This region might also contain a microcomplex that provides an internal forward model for reflexes (
Figure 45
).

Further research on structural-functional similarities and differences between the different zones will provide valuable information for refining our knowledge of the evolutionary features of the cerebellum.

7.
Beyond movements.
It is now clear that the classic view of the cerebellum as a motor center applies to the evolutionarily old flocculonodular lobe and the vermis (zones-A and B). This view applies also to the intermediate part of the cerebellar hemisphere, which is largely devoted to voluntary motor control. However, the intermediate part is involved also in sensory cancellation, which helps movements by removal of sensory perturbation, rather than by modulating the movements themselves. The unique circuit organization for the C
₂
-zone suggests such function (
Figure 48
). In this regard, recall that whales have an enormously developed C
₂
zone associated with a large posterior interpositus nucleus (
Oelschläger and Oelschläger, 2009
). It is an interesting possibility that whales’ particularly developed C
₂
-zone-posterior-interpositus nucleus system is used for sensory cancelation and utilized for echolocation and/or acoustic communication across the ocean (
Chapter 2
, “
Traditional Views of the Cerebellum
,”
Section 2
).

The lateral cerebellar hemisphere (D
₁
- and D
₂
-zones, underlain by the dentate nucleus) is even further remote from the sites for the modulation of movements. Shambes et al. (
1978
) showed that tactile sensory signals reach in abundance the rat cerebellar hemisphere. Gao et al. (
1996
) and Parsons et al. (
1997
) recognized that the human dentate nucleus is activated in association with sensation, but not with movements. The dentate nucleus can be activated by cutaneous stimuli, even when there are no accompanying overt finger movements. On the other hand, finger movements not associated with tactile sensory discrimination produced no dentate activation. These findings suggest that the primary function of the dentate nucleus is to process sensory information involved in motor, perceptual, and cognitive tasks.

This interpretation is not incompatible with the present hypothesis that the dentate nucleus is the final output of the microcomplexes, which represent body schema in the D
₁
-zone associated with the premotor cortex (
Figure 50
). Body schemata possess a continually updating map of the self’s body shape and postures (
Chapter 1
, “
Neuronal Circuitry: The Key to Unlocking the Brain
,”
Section 8
). A body schema is a set of neural representations of the body and bodily functions, based on sensory experience (
Chapter 16
,
Section 3
). One may therefore speculate that the sensory activation of the dentate nucleus is a reflection of the body schema representation in the temporoparietal cortex. A similar consideration may also apply to the control of mental activity, for which mental models or Piaget’s schema are formulated in the temporopariteal cortex
and then copied by microcomplexs in the D
₂
-zone as internal models. Such internal models should consist largely of sensory experiences.

Modeling the cerebellar control of complex motor actions and nonmotor cognitive functions is an endeavor that currently features the use of psychological concepts of body schema, motor schema, Piaget’s schema, and mental models. We must realize that these valuable approaches are all in the realm of cognitive science and as yet, are not fully integrated into the field of neuroscience.

The basal ganglia form a massive network in the deep interior of the cerebrum. With the cerebellum, it constitutes the two largest subcortical stations of motor system, with both operating implicitly. From the well-known characteristic symptoms of lesions in the basal ganglia, such as akinesia in Parkinson’s disease and chorea in Huntington’s disease, the primary function of the basal ganglia may appear to augment stabilization of complex activities in the CNS. “Stabilization augmentation” involves the selection of an activity that best fits the behavioral situation and context, and the suppression of other ongoing CNS activity that would interfere with the desired behavior (
Mink, 1996
). This notion is supported by the recent fMRI study of brain activity of professional and amateur players in a board game named Shogi: activations specific to professionals occurred in the caudate nucleus of the basal ganglia during quick generation of the best next move (
Wan et al., 2011
). This stabilization augmentation should not be confused with the “control augmentation” provided by the cerebellum. These two processes may not be achieved by the same strategy. It is important to realize that augmented stability may make a system less controllable or vice versa. Here we see the demand of the CNS to develop separate devices; the cerebellum for control augmentation and the basal ganglia for stabilization augmentation (
Ito, 1986
).

A mechanism that might underlie the basal ganglia’s selection process is inhibiting—the inhibition that the substantial nigra pars reticulata exerts on a target system of the thalamus and/or other brainstem areas. This disinhibtion is exerted by the caudate nucleus that inhibits the substantial nigra pars reticulata. Selection occurs within the caudate nucleus where various cortical inputs “compete” and certain neurons “win.” The winning caudate neurons are thought to then suppress surrounding caudate neurons in a center (excitation)-surround (inhibition) manner. The result is that only a limited number of the substantia nigra pars reticulata neurons would be inhibited and their target system thereby released from inhibition. This disinhibitory pathway is presumably associated with a side path involving the subthalamic nucleus and the external segment of the globus pallidus, which may
enhance inhibition in the substantia nigra pars reticulata. The interaction of dopaminergic projections from the substantia nigra pars compacta with cortical input in the caudate nucleus is considered to be the primary mechanism underlying this particular learning process. Such a selection-based stabilization mechanism has been investigated in studies on saccadic eye movements (see
Hikosaka et al., 2000
).

The unique symptoms of basal ganglia disorders suggest that its postulated stabilization-by-selection mechanism does indeed regulate the multifaceted movements that occur during posture and locomotion. The mechanism has also been demonstrated for the control of saccadic eye movements. These movements are inhibited normally but, when appropriate, released from inhibition by a cortical mechanism (
Hikosaka and Wurz, 1985
). Although the basal ganglia receive inputs from the entire neocortex, their outputs are directed largely to the prefrontal cortex via the thalamus. This suggests that the basal ganglia contribute to the orderly operation of the prefrontal cortex as a controller by exerting the stabilization-by-selection mechanism against numerous simultaneous, competing, and even conflicting inputs received from the entire neocortex (
Hikosaka et al., 2000
).

Based on remarkable progresses that revealed unique neuronal circuit structures/functions and mechanisms of learning in the basal ganglia, most investigators now accept that the basal ganglia and cerebellum are both anatomically and functionally distinct (
Graybiel, 2005
). Nevertheless, a long-standing question still continues to ask whether the basal ganglia and cerebellum interact directly with each other. Two anatomical pathways have been identified that might come into play in their interactions. (1) The dentate nucleus has a disynaptic projection to the striatum, which is known to operate as an input stage in basal ganglia processing (
Hoshi et al., 2005
). (2) The subthalamic nucleus has a substantial disynaptic projection to the cerebellar cortex (
Bostan et al., 2010
). It remains to be shown just how these pathways function. A useful thought at this stage for the design of future experiments may be that the basal ganglia select a certain repertoire out of many prepared by the cerebellum in the form of differently tuned microcomplexes.

In this monograph, I have emphasized that the cerebellum is on the forefront of brain structures where there has been a convergence of neurobiological analysis and computer modeling and simulation. Needless to say, the cerebellum is also on the forefront of a vast field of brain diseases where neurological, genetic, and molecular analyses have converged (Manto and Pandolfo, 2002). Efforts in these two major directions will help each other. In
Chapter 1
, I emphasized that cyclic decomposition-reconstruction is a fundamental methodology for studying a complex system like the brain. Admittedly, it is difficult if not impossible to apply this with instant success. Rather, the initial model provided by this approach is usually imperfect and it fails to represent the properties of the original system. Examples are now available, however, that through repeated trials using improved technologies, theories, and modeling, the reconstructed model can indeed come closer and closer to the complex system in its full essence. It is obvious that improved neuroscientific techniques and modeling hardware/software are on the immediate horizon. These advantages co-exist with the caution that there are still deficiencies in our basic knowledge of the cerebellum’s neuronal circuits. Sustained neurobiological efforts are required to clarify such issues at both the cellular/molecular level for analyzing single neurons and the systems level for advancing understanding of how neuronal circuits truly work.

Further efforts are also required to proceed along the hierarchical levels of neural control, from reflexes to brainstem/spinal pattern generation, automatic and voluntary movements, motor actions, and cognitive functions. This is a logical and easily understood way for research on the cerebellum to progress and contribute to the advancement of overall brain research. Research on the cerebellum is also advantageous for considering implicit aspects of brain function. The cerebellum controls adaptively numerous reflexes and rhythmical movements, such as breathing and locomotion, at the unconscious level and also enables us to gain skill in the execution of voluntary movements and motor actions without being aware of just how the skill actually improves. Current research on the cerebellum is rapidly advancing our understanding of how this remarkable structure can govern unconsciously the implicit component of our mental activity.

In this overall endeavor, we come to face the crucial question of how knowledge and idealization are represented and processed in neuronal circuits. This issue, like the creation of artificial intelligence, is one of the key challenges for research throughout the remainder of this century.

Abe H, Shima HM, Sekiguchi H, Guo M, Nagao M, Tamura S, Kondo H (1994) Localization of mRNA for protein phosphatase 2A in the brain of adult rats. Mol Brain Res 22:139–143.

Adrian ED (1935) Discharge frequencies in the cerebral and cerebellar cortex. J Physiol (London) 83:33P.