The Cerebellum: Brain for an Implicit Self (2 page)

At a far earlier time, René Descartes (1596–1650) discussed the search for complex mechanisms of the universe and life by using the clock as a metaphor. During his time, this machine was considered the most complex of all the world’s man-made structures. Following Descartes (
1649
), it can be argued, as is prevalent today, that if one can dismantle a clock into its pieces and then successfully reconstruct them into the same functional machine, the precise nature of the clock is revealed. This methodology is still widely applicable when examining an object of unknown nature. It is dissected into simpler pieces, which can be understood, and then an attempt is made to reconstruct a model composed of all the pieces. If this model exhibits all the properties of the original object, it is indeed understood.

The CNS includes the brain (contained within the skull), which weighs 1.3–1.4 kg in humans, and also the spinal cord, which extends into the vertebral canal. On the basis of conventional anatomy, the brain is grossly divided into the brainstem, cerebellum, and cerebrum. The cerebrum is further divided into the basal ganglia, limbic system, and neocortex (
Figure 1
). The cortex of the cerebral hemisphere is further subdivided to 52 areas (
Brodmann, 1909
;
Garey, 1994
) (
Figure 2
). The cerebellar cortex is also subdivided into nearly a hundred areas (see below and Color Plate II). Currently, we know that each of these divisions is composed of characteristic neuronal circuits that consist of numerous neurons of diverse types interconnected with each other via synapses. Moreover, there are even more numerous glial cells and finely branch blood vessels that support and nourish the neurons. The neuronal circuits in each subdivision constitute local networks, which are further integrated to form global neural systems across subdivisions or divisions. Current neuroscience is based on the belief that these networks and systems operate through specific mechanisms and play specific functional roles in the living body.

Figure 1. A sketch of major divisions of the CNS.

Figure 2. Brodmann’s cerebral cortical areas.

The original dotted map published by Brodmann (
1909
) is converted here to outlined areas. (The original color version was provided by courtesy of Mark Dubin:
http://spot.colorado.edu/~dubin/talks/brodmann/brodmann.html
.)

How can we unveil such mechanisms and the functional roles of neuronal circuits? The initial approach was to dissect the brain into experimentally manageable parts. This was the strategy adopted a century ago by Sherrington (1857–1952) and his group. They severed a segment of the cat spinal cord from its upper (and sometimes lower) segments (
Figure 3A
). Freed from the effects of other structures, the severed segments exhibited reflexes with stable, straightforward input-output relationships via the dorsal and ventral roots, which could be subjected to precise scientific investigation.

Figure 3. Sketch of some spinal reflex circuitry.

(A) Schematic of some spinal reflex pathways (modified from
Eccles et al., 1954
). (B) Spinal circuitry drawn by Eccles as based on his group’s intracellular recording data on the recurrent Renshaw cell pathway (Eccles, 1963). In this and subsequent figures, sketches of a single cell and fiber (axon) usually represent groups of such units. A includes muscle spindles and their Ia afferents, two spinal cord segments, spinal motoneurons, Ia inhibitory interneurons, and two opposing muscles. B includes motoneurons and their axons supplying parts of muscle fibers, recurrent motor axon collaterals, Renshaw cells, and other spinal inhibitory interneurons. Abbreviations: ACh, acetylcholine; AS, annulospiral endings; BST, biceps and semitendinosus muscles; E, excitatory synapses: I/IS, inhibitory synapses; L6-L7, lumbar spinal segments; Q, quadriceps muscle; QIa, spindle Ia afferents supplying Q spindles. Symbols: black-filled neurons and their endings, inhibitory; open neurons and their endings, excitatory. This figure is dedicated to a 1963 Nobel Laureate, John Carew Eccles (1903–1997), who was my postdoctoral mentor in Canberra, Australia, from 1959 to 1962. (See
Ito, 1997a
,
2000
;
Stuart and Pierce, 2006
.)

When a neuronal circuit is defined in terms of its gross structure and function, it can then be decomposed into its individual neurons and their dendrites, axons, and synapses, using the currently available technologies of neuroscience. Thereafter, one may try to reconstruct a model of the initial reflex circuit by using the
properties of all its constituent parts. In the process of reconstruction, the mechanistic principle(s) operating in the neuronal circuit may well be revealed.

Sherrington’s group assumed that peripheral stimuli induced excitatory and inhibitory “states” in the spinal centers for various reflex circuits. John Eccles (1903–1997) and his colleagues (e.g.,
Brock et al., 1952
) later identified these as formed by the membrane depolarization and hyperpolarization of spinal motoneurons via excitatory and inhibitory synapses (
Figure 3B
). Hubel and Wiesel (1960) discovered the unique responsiveness of individual neurons to line stimuli in the
visual cortex. They proposed a model of a neuronal circuit to explain how the characteristic responsiveness of “simple” and “complex” cells are formed, using input from concentric receptive fields of the lateral geniculate neurons. These early discoveries marked the start of modern neuroscience.

Neuroscience is now dominated by the effort to decompose neuronal circuits into their cellular and molecular components. Many would argue, however, that reconstructing models of such circuits is equally important in our attempt to comprehend their functional principles (e.g., van Hemmen and Sejnowski, 2006;
Stuart, 2007
; for biology as a whole, see
Noble, 2006
). In the reconstruction process, it is possible to uncover novel principles operating in the original neuronal circuits. Analogies to man-made systems such as computers, control devices, and communication networks have also been helpful, as emphasized in the field of cybernetics by Nobert Wiener (1894–1964).

The circular approach through decomposition and reconstruction provides a general method of fundamental research that features close interactions between experiments and theory (
Figure 4
). Initially, a factual observation of a complex subject may suggest a crude conceptual model, which serves to generate a prediction for a more focused experimental observation. If the prediction turns out to be correct, it supports the crude model, which is then refined to a more accurate conceptual model. This, in turn, can be converted into a substantial computational model, which is reproducible on a computer. Such an advanced model enables us to make further predictions, which can again be tested in even more precise experiments. In this iterative, cyclic development using observation-inspired models, model-based predictions, and experimental testing of the predictions, a model is continuously refined until it accurately simulates the complex subject.

Figure 4. A decomposition/reconstruction cycle.

Research on the CNS starts usually with the experimental dissection of a relatively complex system into its simpler elements. To this end, a system is defined as a CNS unit (e.g., a spinal segment, the pineal body, oculomotor system) while it is undertaking a specific operation. In some cases the system can include peripheral effectors (i.e., glands, muscles). The dissected elements are assorted to construct models of the original system by means of theories and simulations. This circular approach may be based on observation-inspired models, model-based predictions, or experimental testing of a prediction. The model is continuously refined until it accurately simulates the complex system, as symbolized by three trajectories, which represent the first cycle (outer trajectory), an intermediate (middle) cycle, and the most refined (inner) cycle.

A well-known and unique difficulty in research on the CNS arises from its highly hierarchical structure. Comprehension of our current understanding of the brain requires knowledge integrated across several hierarchical levels including molecules, cells, circuits, systems, and behaviors. It seems that ever since organic molecules appeared on earth, these hierarchical levels gradually accumulated through evolution until the human CNS evolved. The above-mentioned decomposition-reconstruction approach can be applied to any two successive levels of the overall hierarchy. For example, a simple neuronal circuit set can be reduced to its component neurons having somata, axons, dendrites, and synapses (
Figure 5
). In turn, these component neurons can be combined to reconstruct a model of the circuit at its original hierarchical level. Next, the component neurons can be further reduced to the lower level of ion channels, receptors, first and second messengers, and various organelles, whose combined properties can provide models of electrical and chemical signaling processes in neurons. Ion channels, receptors, and messenger molecules can be further reduced to an even lower level of proteins and their genes. The latter’s properties can be incorporated into models of the original ion channels and signaling molecules. By this method, the initial simple neural circuit can be linked step by step (not by jumps) to the molecular mechanisms subserving neuronal functions.

Figure 5. The progression of decomposition/reconstruction cycles.

Shown are levels of analysis that extend from gene regulation (-2) to the cellular/molecular- (-1), neuronal- (0), simple circuit- (+1), and finally, complex circuit (+2) level of analysis. Major themes at levels -1 to +1 are also shown.