The Cerebellum: Brain for an Implicit Self (31 page)

TMS has been applied to disrupt briefly ongoing neuronal processes in the cerebellum. In the Miall et al. (
2007
) study, participants viewed a virtual image of a static target in three-dimensional (3D) space ahead of them and started each trial by lifting their right index finger from a start key and moving steadily toward their right. Liquid crystal device (LCD) goggles blocked the view of their hand and of the target as soon at the start key was released. An auditory go cue, 500–1500 milliseconds after trial onset, instructed them to make a rapid upward- and leftward-pointing movement to the virtual target (Color Plate XV). Their index finger had typically moved laterally 10–40 centimeters from its original position when the go cue was delivered. Final positional errors on control trials were small and averaged 4.2 centimeters across all conditions. Thus, participants were normally able to compensate for their initial lateral arm motion and reach the target despite the lack of visual feedback. For a random 50% of trials in each block, three TMS pulses were delivered at 50, 100, and 150 milliseconds within their reaction time after the auditory go cue. Errors in setting the initial direction and final finger position of this target-reaching movement were significantly larger during cerebellar stimulation than during control movements. From this measurement, it was calculated that the reaching movement was planned and initiated, as based on hand position, which was estimated to be 138 milliseconds before the hand actually started to move toward the target (
Miall et al., 2007
).

Throwing a ball.
Human subjects’ ability to accurately throw a ball of clay at a visual target is improved by daily training. It has been shown that when wearing
wedge-prism spectacles, control subjects initially threw in the direction of the prism-bent gaze, but after repeated throws they became adapted and were able to hit the target. Immediately after removal of the prisms, the adapted throw persisted, this being the so-called negative aftereffect. Repeated throws were required to adapt back to the control situation. Patients with lesions in the cerebellum or related structures showed impaired or absent prism adaptation (
Martin et al., 1996a
). Furthermore, training control subjects with the right hand did not generalize to left hand performance, and overhand training seldom generalized to underhand throwing. When two subjects threw with the same hand (right) and the same type of throw (overhand) alternately, with and without prisms, over a period of six weeks, they gradually learned to hit the target on the first throw, with and without prisms. The two gaze-throw calibrations (prism and no-prism) were retained for >27 months (
Martin et al., 1996b
).

The skill in using fingers (digital dexterity) varies over seven ranks along the evolution of mammals to primates (examples in brackets): (1) Fused retrained fingers [horse]; (2) Nonconvergent digits [cat]; (3) Nonprehensile digits [rat]; (4) Nonopposable thumb [marmoset]; (5) Pseudo-opposable thumb [bushbaby]; (6) Power grip [chimp]; (7) Precision grip [man] (
Napier and Napier, 1967
). Attempts have been made to correlate this rank of digital dexterity to the size of the corticospinal tract, but in vain (
Heffner and Masterson, 1975
). Digital dexterity might well be related to the development of the structures involved in the cerebrocerebellar communication loop. The following studies have provided knowledge about neuronal mechanisms for the highest ranks (6 and 7) of digital dexterity.

In grasping, lifting, and replacing an object by hand, humans exhibit subtle grip force control with sufficient yet minimal force, thus minimizing the risk of crushing or inadvertently dropping the object. When human subjects lift small objects using a precision grip between the tips of the fingers and thumb, the ratio between the grip force and the load force (i.e., the vertical lifting force) takes into account friction between the object and the skin. Johansson and Westling (
1987
) provided direct evidence that signals in tactile afferent units are utilized for this purpose. Tactile afferents were readily excited by small but distinct slips between the object and the skin, which were sensed as vibrations of the object. Such adaptation in grip force versus load force coordination should involve the cerebellum, and indeed, cerebellar lesions in stroke patients exhibited slowed grip force development and impaired grip force versus lift force coordination (
Müller and Dichgans, 2002
).
In monkeys trained to perform a grasp-lift-hold task, inactivation by muscimol injection into the anterior interpositus nucleus region produced pronounced dynamic tremor and dysmetric movements of the ipsilateral arm when the animal performed unrestrained reaching and grasping movements. In contrast, no immediate deficits were observed during a 15–20 minute period after muscimol injection in the dentate nucleus, albeit some effects were observed after several hours (
Monzée et al., 2004
). Thus, grasp-lift-hold tasks appear to be controlled primarily by the anterior interpositus nucleus, and not the dentate nucleus.

To examine neural events underlying the preceding type of task, researchers trained monkeys to use a precision grip in order to lift and hold for 2.5 seconds an instrumented object at a fixed height. In some blocks of 20–30 trials, a downward force-pulse perturbation simulating object slip was applied to the object after 1.5 seconds of stationary holding. The animals were required to resist the perturbation to obtain a fruit juice reward. The perturbations produced two types of responses. (1) They elicited invariably a reflex-like, time-locked increase in grip force at latencies between 50 and 100 milliseconds. These persisted as long as the perturbation was applied. (2) When the trials were repeated, a preparatory increase in grip force emerged prior to the onset of the perturbation, and it was extinguished slowly after the perturbation was removed (
Monzée and Smith, 2004
). This preparatory response may represent a learned component of the grasping movement that develops during repeated trials.

During repeated trials of a grasping-lifting-holding task, single cells were recorded in and around the interpositus and dentate nuclei and the overlying cerebellar cortex (Espinoza and Smith, 1990; Dugas and Smith, 1992;
Monzée and Smith, 2004
). Among those nuclear cells whose activity was related to grasping and lifting, the vast majority responded to a downward force-pulse perturbation simulating object slip (see preceding description). The majority of these cells displayed both reflex-like and anticipatory responses to the perturbation. They were confined to the dorsal anterior interpositus nucleus, that is, adjacent to, but not within, the dentate nucleus (
Monzée and Smith, 2004
). The object-slip perturbation also increased powerfully simple-spike discharges in Purkinje cells recorded in the anterior paravermal and lateral cerebellar cortex. The strong responses of these Purkinje cells suggest that the cerebellum participates in corrective responses. Repetitive perturbations elicited preparatory short-latency responses from many of these Purkinje cells (Dugas and Smith, 1992), which might represent preparatory motor responses in cerebellar circuits.

In an fMRI study, control human subjects performing a precision grip with the right arm and hand displayed activity only in the right, anterior, and superior
cerebellum and/or biventer of the left cerebellum (
Kawato et al., 2003
). These cerebellar regions have been implicated in forward models of arm movement (
Chapter 15
). In patients with cerebellar lesions, however, task performance was impaired as revealed by a disruption of temporal coordination between proximal (lifting) muscles and (gripping) fingers. This result suggested that the cerebellum helps coordinate the timing of multijoint movement sequences. Damage to the dentate nucleus or, in particular, its afferent input, leads to an increase in grip force (
Fellows et al., 2001
). Modeling has been carried out on the control process in humans undertaking the preceding types of task, which included variable initial grip apertures and perturbations and variations in object size, location, and orientation. This modeling incorporated slip sensations as error signals in the cerebellum to adapt phasic motor commands to tonic force generators associated with output synergies controlling grip aperture (
Ulloa et al., 2003
,
2010
;
Ulloa and Bullock, 2003
). A computational model has also been constructed that involves the cerebellum in learning anticipatory grip force control by referring to grip force error estimation and sensory input on deformation of the finger tips (
De Gruijl et al., 2009
).

In operant (instrumental) conditioning, humans and animals learn to behave in a specific manner to obtain rewards or avoid punishments. The major mechanism for this process is generally considered to reside in the cerebral cortex. The cerebellum, however, may also be involved, as revealed by an experiment in which a monkey was successfully trained, using operant conditioning, to repeatedly lift a lever in response to visual stimuli. At an early stage in the learning process, when the monkey still lifted the lever randomly, short-latency electrical responses to a light stimulus appeared bilaterally in certain areas of the prefrontal, premotor, and prestriate cortices. These responses became gradually larger as the monkey’s training progressed. When the monkey started to respond to the stimulus with the appropriate movement, premovement potentials appeared in the forelimb motor cortex, and the size of responses in the premotor cortex increased. As the movement became faster and more skillful, late premovement potentials emerged and became even more marked, larger, and steeper in the forelimb motor cortex contralateral to the moving hand. A cerebellar hemispherectomy contralateral to the motor cortex eliminated these potentials. This suggested the participation of the neocerebellum at an advanced stage of the learning process (
Sasaki et al., 1981
;
Sasaki and Gemba, 1982
).

According to a standard definition of “skill” as being an adaptive behavior acquired through practice (
Chen et al., 2005
), operantly conditioned changes in the short-latency (tendon tap) spinal stretch reflex and H-reflex constitute simple motor skills. They can be manipulated to increase or decrease the size of response to an operant conditioning task, as demonstrated in monkeys and rats (
Chen and Wolpaw, 2005
). The two reflexes changed when the testee was exposed to an operant conditioning protocol that gradually decreased (down-conditioning) or increased (up-conditioning). For example, when the H-reflex of the soleus muscle was elicited in freely moving rats by stimulating the posterior tibial nerve, down-conditioning was induced by giving a food reward 200 milliseconds after the H-reflex amplitude became smaller than a criterion value (
Wolpaw and Chen. 2006
). Concerning neuronal mechanisms for this form of adaptation, it has been reported recently that down-conditioning of the H reflex is accompanied by an increase in the number of involved GABAergic interneurons and their synaptic terminals in the spinal cord (
Wang et al., 2006
,
2009
). This may have caused the down-conditioning, but on the other hand, ablation of the interpositus and lateral cerebellar nuclei in down-conditioned rats caused an immediate increase and a delayed increase with an overshoot in the H-reflex. This effect was observed for as long as 150 days. Therefore, the cerebellum appears also to play an essential role in the maintenance of down-conditioning (
Wolpaw and Chen, 2006
). Because a lesion experiment indicated that down-conditioning requires the corticospinal tract and does not require other major descending pathways (
Chen and Wolpaw, 2002
), the cerebellum may use this pathway in the down-conditioning process.

Voluntary control systems differ from reflex control systems in that the former operate with instruction signals generated centrally within the cerebral cortex (
Figure 9C
), whereas the latter are driven primarily by peripheral stimuli. The motor cortex receives instruction signals that designate the content of the movement to be performed. For example, this can be a desired trajectory for an arm movement or even a complex program of movements. The primary motor cortex forms motor command signals and sends them to the brainstem and spinal segmental levels via the corticospinal descending tract. In higher nonhuman primates and humans, part of the primary motor cortex projects corticospinal fibers to directly innervate shoulder, elbow, hand, and finger motoneurons (
Rathelot and Strick, 2009
). On the other hand, the rostral region of the primary motor cortex represents the “old” primary motor cortex, which sends its descending commands to motoneurons indirectly via segmental circuits (see
Lemon, 2008
).

What are the sites for generation of the instruction signals to the primary motor cortex? These should arise from the cortical areas devoted to motor planning and preparation, that is, directly from the anterior cingulate gyrus, supplementary motor area, and premotor area and indirectly from the presupplementary motor area via the supplementary motor area. Electrical stimulation of the supplementary motor area in patients has been shown to induce limb movements and to also evoke sensations such as the urge to perform a movement or anticipation that a movement is about to occur (
Fried et al., 1991
). Stimulation of the ventral bank of the anterior cingulate sulcus evoked an irresistible urge in a patient to grasp something, resulting in exploratory eye movements and a wandering arm (
Kremer et al., 2001
). These findings are in contrast to the finding that TMS of the primary motor cortex does not induce conscious awareness even when it produces a movement (for example, see
Haggard et al., 2002
). An fMRI imaging study revealed that self-paced thumb movements caused cerebral activation to spread from the anterior cingulate gyrus through the supplementary motor area and premotor area to the primary motor and sensory cortices. This cascade in the cerebral cortex occurred temporally in parallel to cerebellar activation, which propagated from lateral to medial parts of the cerebellar cortices (
Hulsmann et al., 2003
). These observations support the view that instruction signals for voluntary movements arise from “higher” cortical regions that include the anterior cingulate gyrus, supplementary motor area, and premotor area.