The Cerebellum: Brain for an Implicit Self (32 page)

Neuronal mechanisms for voluntary movements of the arms, hands, and fingers have been studied using observations on the effects of cerebellar lesions or TMS and brain imaging in monkeys and humans and unit recording from Purkinje cells in monkeys. These studies have been fruitful in revealing some of the basic neuronal mechanisms controlling pointing, reaching, throwing, grasping, and gripping. These are the basis for developing the concept of the internal-model-based control discussed in
Chapter 15
.

For securing well-focused foveal vision, the frontal eye field of the cerebral cortex controls three types of voluntary eye movement; smooth pursuit, saccades, and vergence. An additional eye field is located in the dorsomedial frontal cortex, which appears to play roles in integrating voluntary eye movements with nonoculomotor cortical functions. Four distinct areas of the cerebellum are involved in the three types of eye movements: (1) paraflocculus/flocculus, (2) lobulus petrosus of the paraflocculus, (3) vermal lobules VI/VII, and (4) crus I and crus II (corresponding to HV–HVII). These four areas have different roles in tracking objects moving in three-dimensional space.

The frontal eye field in monkeys is located in Brodmann’s area 8 (see
Figure 2
). Brain imaging and lesion data revealed a remarkable consistency regarding the rostro-caudal and dorso-ventral location of the human frontal eye field, but there was a marked variability along the mediolateral axis, which might challenge the commonly held view of the frontal eye field being located in Brodmann’s area 8 (
Pause, 1996
). Single-unit recording in monkeys and imaging the human brain have shown that the frontal eye field is certainly involved in the execution of eye movements. The field projects to the superior colliculus. The latter’s temporary inactivation (e.g., by the local injection of lidocaine or muscimol) suppressed saccades (
Hanes and Wurtz, 2001
). Note also that the frontal eye field provides extensive feedback connections to the extrastriate visual cortex, thereby suggesting that neuronal activity in the field can influence neural processing in the extrastriate visual cortex (
Schall et al., 1995
).

Using intracortical microstimulation, subregions of the frontal eye field in monkeys have been mapped for their effects on smooth versus saccadic eye movements. This revealed that the saccadic subregion of the frontal eye field is distinct from the subregion for smooth pursuit, a result that was confirmed by single-unit recording (
Tian and Lynch, 1996
). In an fMRI study on human subjects, it was shown that frontal eye field activation during smooth-pursuit performance was smaller than that during saccades. This finding is consistent with the relative size of the two subfields found in humans versus monkeys. Another of the fMRI findings was that the mean location of the smooth pursuit-related frontal eye field in humans was more inferior and lateral than the location of the saccade-related field (
Petit et al., 1997
).

Another single-unit recording study in anaesthetized monkeys showed that frontal eye field neurons have patterns of discharge that differed for saccadic versus pursuit movements. Two types of neurons (I and II) were identified. Type I cells fired during saccades in a specific direction and during the saccade-like, fast phase of nystagmus. These neurons were silent during slow-pursuit movements. In contrast, Type II cells showed steady discharges when the eyes were oriented in a specific direction, and they also discharged during smooth-pursuit movements and the slow phase of nystagmus (
Bizzi, 1968
).

In the frontal eye field, certain neurons signal the location of conspicuous and meaningful stimuli that may be the targets for saccades. Other neurons control saccade production as to whether and when the gaze shifts. The presence of two loosely coupled distinct neural processes, one for visual selection and the other for saccade production, appears to be required for flexible, visually guided behavior involving switching between prosaccade (movement to an indicated target) and antisaccade (in the opposite direction) movements (
Schall, 2002
; see also following sections).

The various patterns of frontal eye field discharge have also been found in the medial superior temporal visual area (MST) and the supplementary eye field (
Fukushima et al., 2004
;
Akao et al., 2005a
,
b
; see following sections). Moreover, it has been shown that a region of the frontal cortex located immediately anterior to the saccade-related frontal eye field region is involved in vergence and ocular accommodation (
Gamlin and Yoon, 2000
). MST and frontal eye field neurons are likely involved conjointly in the initiation of vergence eye movements.

Humans and monkeys use information from their high-acuity retinal fovea to help control voluntary, well-executed binocular eye movements that can track a small moving target. When trained to pursue a target moving in the horizontal plane in a controlled ramp mode, the monkey shows first an initial slow, smooth pursuit; second, a small catch-up saccade; and finally, a postsaccadic smooth pursuit that almost matches the ramp target’s velocity. Feedback is not available for ~100 milliseconds after the onset of such movements due to a delay in visual processing (
Smith et al., 1969
). Hence, the pursuit at the initial 100 milliseconds of the target’s movement operates in a feedforward manner. This requires the cerebellum to intervene, just as it does in other types of eye movement like the VOR, OKR, and OFR (
Chapter 10
, “
Ocular Reflexes
”).

Neuronal circuits for smooth pursuit involve the temporal and frontal lobes of the cerebral cortex and distinct regions of the cerebellum (
Lisberger et al. 1987
;
Keller and Heinen, 1991
; Krauzlis, 2004;
Thier and Ilg, 2005
). The middle temporal (MT) and medial superior temporal (MST) areas in the superior temporal sulcus process visual motion and oculomotor signals that are typically required for pursuit, and these are conveyed to the flocculus/ventral paraflocculus of the cerebellum via the pontine nuclei, primarily through the dorsolateral pontine nucleus. A second cerebro-cerebellar pathway originates in the frontal eye field and continues through the nucleus reticularis tegmenti pontis, which provides outputs exclusively to the cerebellum: in this case lobules VI and VII of the vermis. There is also evidence for the involvement of the cerebellar hemisphere, crus I and crus II (HV–VII), and the lobulus petrosus (see following sections).

The cerebellum was shown to be implicated in the adaptation of smooth pursuit, when it was induced by repeating the target’s movement for 100–300 milliseconds immediately after the onset of a catch-up saccade. Injections of NO absorber or NO synthase inhibitor (that blocks conjunctive LTD) into the subdural space above the paraflocculus-flocculus scarcely affected the velocity of smooth-pursuit but these injections markedly depressed an adaptation of smooth pursuit. This suggested that conjunctive LTD underlies adaptation of the velocity of smooth-pursuit (
Nagao and Kitazawa, 2000
). Another role of the cerebellum was revealed when the target was turned off temporarily, while a monkey was following circular trajectories that were clockwise or counterclockwise. The fact that pursuit was maintained in the absence of the target gave sure evidence of the predictive nature of the smooth pursuit (
Suh et al., 2000
). While a monkey was required to track a target moving along one trajectory selected randomly from four sum-of-sines trajectories (two horizontal, two vertical), Purkinje cells located in the ventral paraflocculus/flocculus discharged simple spikes for 12 milliseconds before eye motion. This time was slightly longer than the 9-millisecond transmission delay between flocculus stimulation and eye motion. This suggests that these Purkinje cells drove pursuit along predictable sum-of-sines trajectories (
Suh et al., 2000
).

Purkinje cells in regions of the monkey’s ventral paraflocculus/flocculus have been shown to receive mossy fiber inputs of vestibular, visual, and oculomotor origins, as recently reviewed (
Lisberger, 2009
). These Purkinje cells exhibit simple-spike responses during the initiation of smooth-pursuit eye movements. During pursuit of sinusoidal target motion, these Purkinje cells discharged complex spikes modulated out-of-phase with the simple-spike firing rate (
Stone and Lisberger, 1990b
). For target motion in the preferred direction, Purkinje cells in the same regions showed a large transient increase in simple-spike firing rate at the onset of pursuit, a smaller but sustained increase during the maintenance of pursuit, and a smooth return to baseline firing at the offset of pursuit (
Krausliz and Lisberger, 1994
). Pukinje cells in the posterior vermis also exhibited smooth-pursuit eye-movement-related activity (Suzuki and Keller, 1988).

In another study on monkeys, ablation of the vermal VI/VII lobules induced changes in the dynamic properties of smooth-pursuit eye movements; during the open-loop period (the first 100 milliseconds), in particular. Changes included a decrease in peak eye acceleration and a decrease in the velocity at the end of the open-loop period. When the pursuit was adapted by halving or doubling eye velocity (i.e., by wearing lenses at ×0.5 and ×2 magnification, respectively), the main pattern of change was a decrease in peak acceleration following the former training and an increase in the duration of peak acceleration following the latter. This adaptive capability was impaired by lesioning vermal lobules VI/VII. These results suggest that lobules VI/VII play a critical role in both the immediate online and the adaptive control of smooth pursuit (
Takagi et al., 2000
).

Three lines of evidence suggest the involvement of the lateral cerebellar hemispheres in smooth-pursuit eye movements: (1) in human patients, unilateral cerebellar lesions impaired initiation of ipsilateral smooth-pursuit eye movements (
Straube et al., 1997
); (2) in monkeys, electrical stimulation of crus I and crus II (HV–VII) and the dentate nuclei evoked smooth-pursuit-like eye movements (
Ron and Robinson, 1973
); and (3) in monkeys, destruction of the ipsilateral unilateral cerebellar hemisphere impaired initial pursuit movements and decreased the velocity of postsaccadic movement. The pursuit’s adaptation was also impaired (
Ohki et al., 2009
).

In monkey cerebellum, the lobulus petrosus is inserted between the dorsal and ventral paraflocculus. This lobule receives inputs from visual system-related pontine nuclei, and it projects to eye-movement-related cerebellar nuclei. Results obtained in monkeys using tracers have shown that the lobulus petrosus and crus I/II of lobule HVII share some of their mossy and climbing fiber inputs, suggesting
that these two areas have similar functional roles in the control of smooth-pursuit movements. In this study, monkeys tracked a moving ramp target using an initially slow eye movement, then a small catch-up saccade, and finally a postsaccadic pursuit that nearly matched the velocity of the moving ramp. After unilateral lesioning of the lobulus petrosus (by local injections of ibotenic acid), no consistent changes were seen in the amplitude of the catch-up saccades, but the velocity of postsaccadic pursuits in the ipsilateral and downward directions decreased by 20%–40%. These deficits lasted for at least one month, after which some recovery was observed (
Xiong et al., 2010
). These findings suggest the involvement of the lobulus petrosus in the monkey’s control of smooth pursuit.

In yet another study, monkeys were trained to pursue a moving target either with their head moving freely in the horizontal plane (with their eyes voluntarily held stationary in their orbits) or with their freely moving eyes (with their head voluntarily held stationary in space). Neurons in the rostral portions of the nucleus reticularis tegmenti pontis were found to encode head-pursuit velocity and head-pursuit acceleration. The vast majority of the tested neurons exhibited responses to both head and eye pursuits. As such, the neurons’ discharge contained a provisional gaze-pursuit signal (
Suzuki et al., 2009
). The frontal eye field projects to the nucleus reticularis tegementi pontis, which, in turn, projects to the cerebellar nuclei. The nucleus reticularis tegmenti pontis also receives inputs from the superior colliculi and pretectal midbrain areas (
Gamlin and Clarke, 1995
). Therefore, rostral portions of the nucleus reticularis tegmenti pontis are a strong candidate for the source of an active head-pursuit signal that projects to the cerebellum: that is, specifically to the Purkinje cells in vermal lobules VI and VII that have target-velocity and gaze-velocity-encoding properties.

Concerning neuronal mechanisms of smooth pursuit, there is as yet no consensus among researchers. Lisberger (
1994
) assumed a strong positive feedback loop from the vestibular nucleus to the cerebellar cortex (as in
Figure 39B
). This structure was thought to play an essential role in reproducing both VOR adaptation and smooth pursuit. On the other hand, Tabata et al. (
2002
) assumed that the pursuit driving command was generated and maintained in the MST area (as in
Figure 39C
). There seems to still be much to work out in the attempts to model neuronal mechanisms of smooth pursuit.

Saccades can be evoked not only reflexively via the superior colliculus and the brainstem circuit (
Chapter 10
), but also voluntarily via the frontal eye field of the
cerebral neocortex. Lesioning of the frontal eye field causes deficits in (1) generating saccades to briefly presented targets, (2) the production of saccades to two or more sequentially presented targets, and (3) the selection of simultaneously presented targets. Thus, in voluntary saccades, the frontal eye field acts as the controller and the brainstem saccadic system and eyeballs act as the controlled object. In a typical short-term saccadic adaptation protocol, the target moved midflight during the saccade either toward (gain-down) or away (gain-up) from initial fixation, and so caused the saccade to complete with an endpoint error. In the gain-down paradigm, the adapted saccades showed reduced peak velocities, accelerations, and decelerations, and increased durations compared with a control saccade of equal amplitude. (
Ethier et al., 2008
). The properties required for a model involved in voluntary saccades will be discussed later in
Chapter 15
, “
Internal Models for Voluntary Motor Control
.”