The Cerebellum: Brain for an Implicit Self (24 page)

At first glance the OFR operates by feedback, but in reality it is a feedforward mechanism. This is known because of its long loop time (50 ms; i.e., long for visual information processing across the retina and MT/MST areas) in response to 100 milliseconds ramp changes in the visual scene. One relevant study in monkeys, which was designed to induce OFR adaptation, involved application of double-ramp sequences: the first ramp to initiate the OFR and the second to apply visual errors that evoked adaptive mechanisms in the OFR. Both ramps lasted 150 milliseconds and were delivered at different speeds and directions (speed steps and direction steps, respectively). Repeated exposure to increasing versus decreasing speed steps increased and decreased, respectively, the OFR (
Miles and Kawano, 1986
). Another study analyzed simple spike discharges of Purkinje cells using the multiple linear regression technique for system identification. It was found that dorsolateral pontine nucleus neurons, the origin of mossy fiber inputs to the ventral paraflocculus, encoded some selective aspects of visual stimuli (
Takemura et al., 2001
). It was found also that simple spike activity in ventral paraflocculus Purkinje cells encoded eye movements in terms of their acceleration, velocity, and eye position, thereby representing inverse dynamics of the actual eyeball movements (
Shidara et al., 1993
). The meaning of these findings will be considered in
Chapter 12
.

Whereas the VOR in rabbits is adaptively controlled by the flocculus, three specific types of ocular reflexes (VOR, OKR, and OFR) in monkeys are under the influence of a cerebellar area composed of the flocculus and ventral paraflocculus. How these reflexes are differentially represented in this area and integrated to perform purposeful eye movements is still a matter of conjecture. One difficulty is that the flocculus and ventral paraflocculus exhibit a particularly large species variation in folial morphology (
Voogd, 2004
). A practical solution for arguments of folial structures is to ignore them and lump these areas as one floccular complex, as adopted in physiological mapping in monkeys (
Lisberger, 2009
). For a radical solution, nevertheless, precise mapping is required for topographical demarcation of the cerebellar cortex in association with precise knowledge of the relevant neuronal circuits and functions. For example, the area defined as the proper flocculus receives major mossy fiber inputs from (1) the primary vestibular nerve, (2) vestibular nuclei, (3) the NRTD, and (4) the central part of the mesencephalic reticular formation. In contrast, the area defined as the ventral paraflocculus receives major mossy fiber inputs from pontine nuclei and NRTP (
Gerrits and Voogd, 1989
;
Glickstein et al., 1994
; Nagao et al., 1997a). It is apparent that the former area mediates the VOR/OKR, and the latter the OFR. On the other hand, Purkinje cells in the flocculus project to the medial and ventrolateral parts of the medial vestibular nucleus, superior vestibular nucleus, and the y group, which contain VOR relay neurons. Purkinje cells in the ventral paraflocculus also project to the medial and ventral parts of the medial vestibular nucleus, superior vestibular nucleus, and y group (
Balaban et al., 1981
; Nagao et al., 1997b). There is also a projection to the caudoventral part of the posterior interpositus and dentate nuclei. Hence, it is possible that the OFR is also mediated by VOR relay neurons. These findings are the basis for the assumption that the VOR, OKR, and OFR are integrated into a multi-input system sharing the controller and the controlled object (
Figure 9A
).

A saccade is a quick, simultaneous movement of both eyes in the same direction to catch a visual target by small foveal areas of the retinae for high-acuity vision. It is part of the overall orienting response to the sudden onset of a novel or behaviorally interesting stimulus, which orients the eyes, external ears, head, and/or body toward the source of the stimulus. Saccades are the fastest movements produced by the human body. (The peak angular speed of the monkey eye during a saccade
can reach 1,000 degrees/second). Saccades to an unexpected stimulus normally take ~200 milliseconds to initiate, and they last for ~20–200 milliseconds. Under certain laboratory conditions, the latency of saccade production can be cut nearly in half, that is, to ~100 milliseconds (for express saccades).

The brainstem neuronal circuit for the saccade generator has been dissected in detail (see
Moschovakis, 1996
;
Scudder et al., 2002
;
Ramat et al., 2006
;
Williams and Hilmas, 2010
). In brief, omnipause neurons in the superior colliculus are activated by a visual stimulus and, in turn, activate excitatory burst neurons located mainly in the caudal pontine reticular formation and inhibitory burst neurons in the medullary reticular formation (
Strassman et al., 1986a
,
b
). For horizontal saccades, excitatory burst neurons activate ipsilateral abducens motoneurons (supplying the ipsilateral lateral rectus muscle) and also contralateral oculomotor neurons (supplying the contralateral medial rectus muscle), the latter effect being mediated by intranuclear neurons in the abducens nucleus. Inhibitory burst neurons inhibit contralateral abducens motoneurons and, at the same time, ipsilateral oculomotor neurons via intranuclear neurons. An excitatory/inhibitory neuron pair thus provides reciprocal innervation to appropriate extraocular muscles to make a saccade in one horizontal direction. Some of these neurons are included in a model of cerebellar control of the saccade generator (see
Figure 40
).

A neural mechanism in the superior colliculus directs a saccade to a target in the visual field. Electrical microstimulation in a deep layer of the superior colliculus evokes a saccade, whose amplitude and direction are functions of the site of stimulation and independent of its intensity and frequency. As the stimulating site in the superior colliculus is moved caudally and medially, respectively, the amplitude of the saccade increases and the direction of the saccade shifts from downward to upward (for a two-dimensional motor map of saccades, see
Robinson, 1972
;
Sparks and Nelson, 1987
). The timing of the onset and cessation of the discharge of excitatory and inhibitory burst neurons varies according to the site of activation in the superior colliculus. The amplitude and direction of the saccade so induced is appropriate for a foveal catching of the target in the visual field. A group of neurons called long-lead burst neurons has been found to be concentrated in the rostral pons. These neurons are characterized by low-level, unstructured activity preceding (by ~30 milliseconds) the intense peri-saccadic burst. Long-lead burst neurons appear to play multiple roles in saccades; for example, some of them may mediate a trigger to suppress omnipause neurons’ discharge (see previous description) during small saccades (
Kaneko, 2006
).

The involvement of the posterior vermis (lobules VI and VII) in saccade control has been determined by the observations that (1) local electrical stimulation of
this area induced saccadic eye movements, and (2) this effect was lost when Purkinje cells in the same area were destroyed by a local kainate injection (
Noda and Fujikado, 1987
). Purkinje cells of the posterior vermis project their axons to the posterior part of the fastigial nucleus, which, in turn, projects to the previously described saccade-generator circuit in the lower brainstem. Saccadic eye movements are so rapid that their precision control cannot be based on visual feedback. Rather, cerebellar mechanisms are required for this purpose. For example, saccades were shown to become grossly inaccurate (hypometria) three days after ablation of the monkey cerebellum. Three or more months later, the average saccade amplitude was recovered near completely, but it remained more variable than before the ablation (
Barash et al., 1999
). When the GABA agonist, muscimol, was injected unilaterally or bilaterally into the caudal fastigial nucleus, the saccades became hyper- or hypometric and their trajectories and end points became more variable (
Robinson et al., 1993
). Posterior vermis lobules VI/VII and the caudal fastigial nucleus are therefore presumed to act on the saccade generator in the lower brainstem to make saccades more consistent and accurate.

The adaptive control function of the cerebellum for saccades has been tested by a paradigm called “saccadic adaptation.” In control trials, a first jump of a visual target by 15° caused an accurate saccade. In adaptation trials, while a similarly induced saccadic movement was under way, the target was displaced from 15° to 20°, the resultant eye position ending 5° short of the target. After 1,000 consecutive adaptation trials, eye position overshot the original 15° position of the target and ended closer to the 20° position. Lesioning of the posterior vermis abolished permanently this saccadic adaptation (
McLaughlin, 1967
;
Barash et al., 1999
).

Close connections of the cerebellum to the saccade generator have also been demonstrated by showing that a significant proportion of mossy fibers carries a signal very similar to excitatory burst neuron activity in the saccade generator (
Ohtsuka and Noda, 1992
). Moreover, HRP injected into the posterior vermis labeled, among many other brainstem structures, the paramedian pontine reticular formation (
Yamada and Noda, 1987
), which contains excitatory and inhibitory burst neurons (see previous description). The population response of a large group of Purkinje cells in the posterior vermis was shown to provide a precise temporal signature of saccade onset and offset. Modeling revealed that changing the relative contributions of individual Purkinje cells (i.e., by having them discharge at different times throughout the saccade) translated directly into changes in the amplitude of the saccade (
Thier et al., 2000
; Kojima et al., 2010).

The IO plays a role in conveying an error signal in the current hypothesis of cerebellar learning (
Chapters 3
and
9
). The pathway that appears to convey such errors in saccadic adaptation was found to be located in the monkey midbrain tegmentum. Its weak electrical stimulation delivered ~200 milliseconds after a saccade in one horizontal direction produced progressively more marked changes in saccade gain, this being similar to changes induced by adaptation to real visual errors (
Kojima et al., 2007
). Even though the anatomy of this pathway has not yet been identified, a very likely candidate is the collicular output fibers that project to subnucleus b of the medial accessory nucleus of the inferior olive (MAO). This structure has been shown to project, in turn, to the posterior vermis (
Yamada and Noda, 1987
). A computer simulation using these data reproduced the adaptation of saccadic gain to repeated presentations of dual-step visual targets (see
Chapter 12
).

Neuronal circuits for the four types of ocular reflex (VOR, OKR, OFR, and saccades) have been dissected in terms of both their structure and function. In addition, the roles and mechanisms of the cerebellum in their elaboration have been studied in detail. The findings to this point provide a firm basis for clarifying several adaptive control mechanisms of the cerebellum in
Chapter 12
.

In this chapter we will consider several types of somatic and autonomic reflexes whose operation is improved when modulated by the cerebellum. The prototypical adaptive cerebellar control of these reflexes is similar to that for the ocular reflexes analyzed in
Chapter 10
, “
Ocular Reflexes
.” In particular, eye-blink conditioning, together with VOR/OKR adaptation, provide excellent examples of cerebellar function.

A stretch reflex causes the contraction of a muscle in response to its stretch. Interestingly, its mediators are quite complex. First, three types of motoneurons are involved. α-motoneurons innervate extrafusal muscle fibers, whereas γ-motoneurons innervate intrafusal muscle fibers. In addition, β-motoneurons innervate both types of fibers. Second, three types of muscle afferents come into play (Ia, Ib, and spindle II afferents). Ia and Ib afferents are the fastest conducting myelinated axons, and they are subdivided by a combination of their physiological responsiveness to brief muscle stretch and conduction velocity. Spindle II afferents are the muscle spindle component of the group II afferent population (
Matthews, 2010
). Third, spinal interneurons come into play during long-latency components of the stretch reflex.

Group Ia afferents arise from annulospiral endings in the middle area of muscle spindles. They make direct excitatory connections with the α-motoneurons of that muscle and its synergists (they also have disynaptic inhibitory connections with antagonist α-motoneurons that are discussed later and several other connections not covered in this chapter). Hence, the impulse signals of stretch-activated Ia afferents from a muscle may, under some conditions, induce
monosynaptic excitation of α-motoneurons that, in turn, cause contraction of that muscle; that is, the short-latency Ia (also called “tendon tap”) stretch reflex, as it may be called, can provide a negative feedback control of muscle length (length feedback; see
Nichols and Ross, 2009
) (
Figure 31A
). In several vertebrates, including the human, this reflex can be demonstrated (again under certain conditions) by a brief tendon tap (stretch) or in the form of an H reflex, which results from applying an electric shock to largely the group Ia axons of a peripheral muscle nerve (
Pierrot-Deseilligny and Burke, 2005
).