Learning and maintaining saccadic accuracy: a model of brainstem-cerebellar interactions

Multiple behavior-specific cancellation signals contribute to suppressing predictable sensory reafference in a cerebellum-like structure

Movement induces sensory stimulation of an animal's own sensory receptors, termed reafference. With a few exceptions, notably vestibular and proprioception, this reafference is unwanted sensory noise and must be selectively filtered in order to detect relevant external sensory signals. In the cerebellum-like electrosensory nucleus of elasmobranch fish, an adaptive filter preserves novel signals by generating cancellation signals that suppress predictable reafference. A parallel fiber network supplies the principal Purkinje-like neurons (called ascending efferent neurons, AENs) with behavior-associated internal reference signals, including motor corollary discharge and sensory feedback, from which predictive cancellation signals are formed. How distinct behavior-specific cancellation signals interact within AENs when multiple behaviors co-occur and produce complex, changing patterns of reafference is unknown. Here, we show that when multiple streams of internal reference signals are available, cancellation signals form that are specific to parallel fiber inputs temporally correlated with, and therefore predictive of, sensory reafference. A single AEN has the capacity to form more than one cancellation signal, and AENs form multiple cancellation signals simultaneously and modify them independently during co-occurring behaviors. Cancellation signals update incrementally during continuous behaviors, as well as episodic bouts of behavior that last minutes to hours. Finally, individual AENs, independently of their neighbors, form unique AEN-specific cancellation signals that depend on the particular sensory reafferent input it receives. Together, these results demonstrate the capacity of the adaptive filter to utilize multiple cancellation signals to suppress dynamic patterns of reafference arising from complex co-occurring and intermittently performed behaviors.

Cerebellar adaptive mechanisms explain the optimal control of saccadic eye movements

Cerebellar synaptic plasticity is vital for adaptability and fine tuning of goal-directed movements. The perceived sensory errors between desired and actual movement outcomes are commonly considered to induce plasticity in the cerebellar synapses, with an objective to improve desirability of the executed movements. In rapid goal-directed eye movements called saccades, the only available sensory feedback is the direction of reaching error information received only at end of the movement. Moreover, this sensory error dependent plasticity can only improve the accuracy of the movements, while ignoring other essential characteristics such as reaching in minimum-time. In this work we propose a rate based, cerebellum inspired adaptive filter model to address refinement of both accuracy and movement-time of saccades. We use optimal control approach in conjunction with information constraints posed by the cerebellum to derive bio-plausible supervised plasticity rules. We implement and validate this bio-inspired scheme on a humanoid robot. We found out that, separate plasticity mechanisms in the model cerebellum separately control accuracy and movement-time. These plasticity mechanisms ensure that optimal saccades are produced by just receiving the direction of end reaching error as an evaluative signal. Furthermore, the model emulates encoding in the cerebellum of movement kinematics as observed in biological experiments.

Microsaccades inhibition triggered by a repetitive visual distractor is not subject to habituation: Implications for the programming of reflexive saccades

The oculomotor capture triggered by a peripheral onset is subject to habituation, a basic form of learning consisting in a response decrement toward a repeatedly presented stimulus. However, it is unclear whether habituation of reflexive saccades takes place at the saccadic programming or execution stage (or both). To address this issue, we exploited the fact that during fixation the programming of a reflexive saccade exerts a robust but short-lasting phasic inhibition in the absolute microsaccadic frequency. Hence, if habituation of reflexive saccades occurs at the programming stage, then this should also affect the microsaccadic frequency, with a progressive reduction of the inhibitory phase. Conversely, if habituation occurs only at the later stage of saccade execution, the no change in the microsaccadic pattern is expected. Participants were repeatedly exposed to a peripheral onset distractor, and when eye movements were allowed, we replicated the oculomotor capture habituation. Crucially, however, when fixation was maintained the microsaccadic response did not change as exposure to the onset progressed, suggesting that habituation of reflexive saccades does not take place at the programming stage in the superior colliculus (SC), but at the later stage of saccade execution in the brainstem, where the competition between different saccades might be resolved. This scenario challenges one of the main assumptions of the competitive integration model for oculomotor control, which assumes that competition between exogenous and endogenous saccade programs occurs in the (SC). Our results and interpretation are instead in agreement with neurophysiological studies in non-human primates showing that saccadic adaption, another form of oculomotor plasticity, takes place downstream from the SC.

Integrating Brain and Biomechanical Models-A New Paradigm for Understanding Neuro-muscular Control

To date, realistic models of how the central nervous system governs behavior have been restricted in scope to the brain, brainstem or spinal column, as if these existed as disembodied organs. Further, the model is often exercised in relation to an in vivo physiological experiment with input comprising an impulse, a periodic signal or constant activation, and output as a pattern of neural activity in one or more neural populations. Any link to behavior is inferred only indirectly via these activity patterns. We argue that to discover the principles of operation of neural systems, it is necessary to express their behavior in terms of physical movements of a realistic motor system, and to supply inputs that mimic sensory experience. To do this with confidence, we must connect our brain models to neuro-muscular models and provide relevant visual and proprioceptive feedback signals, thereby closing the loop of the simulation. This paper describes an effort to develop just such an integrated brain and biomechanical system using a number of pre-existing models. It describes a model of the saccadic oculomotor system incorporating a neuromuscular model of the eye and its six extraocular muscles. The position of the eye determines how illumination of a retinotopic input population projects information about the location of a saccade target into the system. A pre-existing saccadic burst generator model was incorporated into the system, which generated motoneuron activity patterns suitable for driving the biomechanical eye. The model was demonstrated to make accurate saccades to a target luminance under a set of environmental constraints. Challenges encountered in the development of this model showed the importance of this integrated modeling approach. Thus, we exposed shortcomings in individual model components which were only apparent when these were supplied with the more plausible inputs available in a closed loop design. Consequently we were able to suggest missing functionality which the system would require to reproduce more realistic behavior. The construction of such closed-loop animal models constitutes a new paradigm of computational neurobehavior and promises a more thoroughgoing approach to our understanding of the brain's function as a controller for movement and behavior.

New supervised learning theory applied to cerebellar modeling for suppression of variability of saccade end points

A new supervised learning theory is proposed for a hierarchical neural network with a single hidden layer of threshold units, which can approximate any continuous transformation, and applied to a cerebellar function to suppress the end-point variability of saccades. In motor systems, feedback control can reduce noise effects if the noise is added in a pathway from a motor center to a peripheral effector; however, it cannot reduce noise effects if the noise is generated in the motor center itself: a new control scheme is necessary for such noise. The cerebellar cortex is well known as a supervised learning system, and a novel theory of cerebellar cortical function developed in this study can explain the capability of the cerebellum to feedforwardly reduce noise effects, such as end-point variability of saccades. This theory assumes that a Golgi-granule cell system can encode the strength of a mossy fiber input as the state of neuronal activity of parallel fibers. By combining these parallel fiber signals with appropriate connection weights to produce a Purkinje cell output, an arbitrary continuous input-output relationship can be obtained. By incorporating such flexible computation and learning ability in a process of saccadic gain adaptation, a new control scheme in which the cerebellar cortex feedforwardly suppresses the end-point variability when it detects a variation in saccadic commands can be devised. Computer simulation confirmed the efficiency of such learning and showed a reduction in the variability of saccadic end points, similar to results obtained from experimental data.

Adaptive filters and internal models: multilevel description of cerebellar function

Cerebellar function is increasingly discussed in terms of engineering schemes for motor control and signal processing that involve internal models. To address the relation between the cerebellum and internal models, we adopt the chip metaphor that has been used to represent the combination of a homogeneous cerebellar cortical microcircuit with individual microzones having unique external connections. This metaphor indicates that identifying the function of a particular cerebellar chip requires knowledge of both the general microcircuit algorithm and the chip's individual connections. Here we use a popular candidate algorithm as embodied in the adaptive filter, which learns to decorrelate its inputs from a reference ('teaching', 'error') signal. This algorithm is computationally powerful enough to be used in a very wide variety of engineering applications. However, the crucial issue is whether the external connectivity required by such applications can be implemented biologically. We argue that some applications appear to be in principle biologically implausible: these include the Smith predictor and Kalman filter (for state estimation), and the feedback-error-learning scheme for adaptive inverse control. However, even for plausible schemes, such as forward models for noise cancellation and novelty-detection, and the recurrent architecture for adaptive inverse control, there is unlikely to be a simple mapping between microzone function and internal model structure. This initial analysis suggests that cerebellar involvement in particular behaviours is therefore unlikely to have a neat classification into categories such as 'forward model'. It is more likely that cerebellar microzones learn a task-specific adaptive-filter operation which combines a number of signal-processing roles.

An internal model architecture for novelty detection: implications for cerebellar and collicular roles in sensory processing

The cerebellum is thought to implement internal models for sensory prediction, but details of the underlying circuitry are currently obscure. We therefore investigated a specific example of internal-model based sensory prediction, namely detection of whisker contacts during whisking. Inputs from the vibrissae in rats can be affected by signals generated by whisker movement, a phenomenon also observable in whisking robots. Robot novelty-detection can be improved by adaptive noise-cancellation, in which an adaptive filter learns a forward model of the whisker plant that allows the sensory effects of whisking to be predicted and thus subtracted from the noisy sensory input. However, the forward model only uses information from an efference copy of the whisking commands. Here we show that the addition of sensory information from the whiskers allows the adaptive filter to learn a more complex internal model that performs more robustly than the forward model, particularly when the whisking-induced interference has a periodic structure. We then propose a neural equivalent of the circuitry required for adaptive novelty-detection in the robot, in which the role of the adaptive filter is carried out by the cerebellum, with the comparison of its output (an estimate of the self-induced interference) and the original vibrissal signal occurring in the superior colliculus, a structure noted for its central role in novelty detection. This proposal makes a specific prediction concerning the whisker-related functions of a region in cerebellar cortical zone A₂ that in rats receives climbing fibre input from the superior colliculus (via the inferior olive). This region has not been observed in non-whisking animals such as cats and primates, and its functional role in vibrissal processing has hitherto remained mysterious. Further investigation of this system may throw light on how cerebellar-based internal models could be used in broader sensory, motor and cognitive contexts.

Central syntropic effects elicited by trigeminal proprioceptive equilibrium in Alzheimer's disease: a case report

Introduction

The presented patient, affected by Alzheimer’s disease, underwent neuropsychological evaluation and functional magnetic resonance imaging investigation under occlusal proprioceptive un-balance and re-balance conditions. Saccadic and pupillometric video-oculographic examinations were performed in order to detect connected trigeminal proprioceptive motor patterns able to interfere with reticular formation cerebellum functions linked to visual and procedural processes prematurely altered in Alzheimer’s disease.

Case presentation

A 66-year-old Caucasian man, affected by Alzheimer’s disease and with a neuropsychological evaluation issued by the Alzheimer’s Evaluation Unit, underwent an electromyographic investigation of the masseter muscles in order to assess their functional balance. The patient showed a bilateral lack of all inferior molars. The extreme myoelectric asymmetry in dental occlusion suggested the rebalancing of masseter muscular functions through concurrent transcutaneous stimulation of the trigeminal nerve supramandibular and submandibular motor branches. The above-mentioned method allows detection of symmetric craniomandibular muscular relation that can be kept constant through the use of a cusp bite modeled on the inferior dental arch, called orthotic-syntropic bite. A few days later, the patient underwent a new neuropsychological investigation, together with a functional magnetic resonance imaging study, and saccadic, pupillometric video-oculographic examinations in occlusal un-balance and re-balance conditions.

Conclusions

Comparative data analysis has shown that a re-balanced occlusal condition can improve a patient’s cognitive-attentive functions. Moreover, the saccadic and pupillometric video-oculographic investigations have proven useful both in analyzing reticulo-cerebellar subcortical systems, prematurely altered in Alzheimer’s disease, and in implementing neurological evaluations.

The relative importance of retinal error and prediction in saccadic adaptation

When saccades systematically miss their visual target, their amplitude adjusts, causing the position errors to be progressively reduced. Conventionally, this adaptation is viewed as driven by retinal error (the distance between primary saccade endpoint and visual target). Recent work suggests that the oculomotor system is informed about where the eye lands; thus not all "retinal error" is unexpected. The present study compared two error signals that may drive saccade adaptation: retinal error and prediction error (the difference between predicted and actual postsaccadic images). Subjects made saccades to a visual target in two successive sessions. In the first session, the target was extinguished during saccade execution if the amplitude was smaller (or, in other experiments, greater) than the running median, thereby modifying the average retinal error subjects experienced without moving the target during the saccade as in conventional adaptation paradigms. In the second session, targets were extinguished at the start of saccades and turned back on at a position that reproduced the trial-by-trial retinal error recorded in the first session. Despite the retinal error in the first and second sessions having been identical, adaptation was severalfold greater in the second session, when the predicted target position had been changed. These results argue that the eye knows where it lands and where it expects the target to be, and that deviations from this prediction drive saccade adaptation more strongly than retinal error alone.

Saccade adaptation as a model of learning in voluntary movements

Motor learning ensures the accuracy of our daily movements. However, we know relatively little about its mechanisms, particularly for voluntary movements. Saccadic eye movements serve to bring the image of a visual target precisely onto the fovea. Their accuracy is maintained not by on-line sensory feedback but by a learning mechanism, called saccade adaptation. Recent studies on saccade adaptation have provided valuable additions to our knowledge of motor learning. This review summarizes what we know about the characteristics and neural mechanisms of saccade adaptation, emphasizing recent findings and new ideas. Long-term adaptation, distinct from its short-term counterpart, seems to be present in the saccadic system. Accumulating evidence indicates the involvement of the oculomotor cerebellar vermis as a learning site. The superior colliculus is now suggested not only to generate saccade commands but also to issue driving signals for motor learning. These and other significant contributions have advanced our understanding of saccade adaptation and motor learning in general.

The influence of the consistency of postsaccadic visual errors on saccadic adaptation

The saccadic system is a prime example of motor control without continuous visual feedback. These systems suffer from a strong vulnerability against disturbances. The mechanism of saccadic adaptation allows adjustment of saccades to alterations arising not only from anatomical changes but also from external changes. The weighting of errors according to their reliability provides a strong benefit for an optimized control system. Thus the consistency of visual error should influence the characteristics of adaptation. In the typical adaptation paradigm a visual error is introduced by stepping the target during the saccade by a given amount. In this paradigm, the retinal error varies with the accuracy of the saccade and the step size. To study the influence of error consistency we use a variant of the adaptation paradigm which allows to specify a constant error size. Intrasaccadic target step sizes were calculated with respect to the predicted landing position of each individual saccade. The consistency of the visual error was varied by introducing different levels of noise to the intrasaccadic target step. Different mean intrasaccadic target step sizes were examined: positive target step, negative target step, and a condition in which the mean of the error distribution was clamped to the fovea. In all three conditions saccadic adaptation was strongest when the error was consistent and became weaker as the error became more variable. These results show that saccadic adaptation takes not only the average error but also the consistency of the error into account.

Representation recovers information

Early agreement within cognitive science on the topic of representation has now given way to a combination of positions. Some question the significance of representation in cognition. Others continue to argue in favor, but the case has not been demonstrated in any formal way. The present paper sets out a framework in which the value of representation use can be mathematically measured, albeit in a broadly sensory context rather than a specifically cognitive one. Key to the approach is the use of Bayesian networks for modeling the distal dimension of sensory processes. More relevant to cognitive science is the theoretical result obtained, which is that a certain type of representational architecture is necessary for achievement of sensory efficiency. While exhibiting few of the characteristics of traditional, symbolic encoding, this architecture corresponds quite closely to the forms of embedded representation now being explored in some embedded/embodied approaches. It becomes meaningful to view that type of representation use as a form of information recovery. A formal basis then exists for viewing representation not so much as the substrate of reasoning and thought, but rather as a general medium for efficient, interpretive processing.

Learning signals from the superior colliculus for adaptation of saccadic eye movements in the monkey

Vital to motor learning is information about movement error. Using this information, the brain creates neural learning signals that instruct a plasticity mechanism to produce appropriate behavioral learning. Little is known, however, about brain structures that generate learning signals for voluntary movements. Here we show that signals from the superior colliculus (SC) can drive learning in saccadic eye movements in the monkey. Electrical stimulation of the SC deeper layers, subthreshold for evoking saccades, was applied immediately (approximately 60 ms) after the end of horizontal saccades in one or both directions. The target disappeared during saccades and remained invisible for 1 s to eliminate effects of postsaccadic visual information. Repetitive pairing of saccades with SC stimulation produced a marked, two-dimensional shift in movement endpoint relative to the target location. The elicited endpoint shift took a gradual, approximately exponential course over several hundred saccades as in visually induced saccade adaptation. The shift in movement endpoint remained nearly unchanged after stimulation was discontinued, indicating involvement of neuronal plasticity. When both rightward and leftward saccades were paired with stimulation, their endpoints shifted in similar directions. The endpoint shift was directed contralaterally to the stimulated SC. The direction and size of the endpoint shift depended on the stimulation site in the SC. We propose that the SC, a brainstem structure long known to be crucial for saccade execution, is involved in motor learning and sends signals that dictate the direction of adaptive shift in saccade endpoint.

Complex spike activity of purkinje cells in the oculomotor vermis during behavioral adaptation of monkey saccades

Throughout life, the oculomotor system can correct itself when saccadic eye movements become inaccurate. This adaptation mechanism can be engaged in the laboratory by displacing the target when the saccade toward it is in flight. Forward and backward target displacements cause gradual increases and decreases in saccade amplitude, respectively. Equipped with this paradigm, we asked whether Purkinje cells (P-cells) in the vermis of the oculomotor cerebellum, lobules VIc and VII, changed their complex spike (CS) discharge during the behavioral adaptation of horizontal saccades. We tested the hypothesis that CS activity would change only when a targeting saccade caused an error in eye position relative to the target, i.e., during the error interval between the primary and corrective saccades. We examined only those P-cells whose simple spike activity exhibited either a burst or pause with saccades in several directions. Approximately 80% of such P-cells exhibited an increase in CS activity during the error interval when the adaptation paradigm imposed horizontal eye-position errors in one direction and a decrease in activity for errors in the other. As adaptation progressed and errors were reduced, there was no consistent change in the CS activity. These data suggest that the CS activity of P-cells in the oculomotor vermis signals the direction but not the magnitude of eye-position error during saccade adaptation. Our results are consistent with cerebellar learning models that have been proposed to explain adaptation of the vestibulo-ocular reflex so similar mechanisms may also underlie plasticity of this precision voluntary oculomotor behavior.

From brainstem to cortex: computational models of saccade generation circuitry

The brain circuitry of saccadic eye movements, from brainstem to cortex, has been extensively studied during the last 30 years. The wealth of data gathered allowed the conception of numerous computational models. These models proposed descriptions of the putative mechanisms generating this data, and, in turn, made predictions and helped to plan new experiments. In this article, we review the computational models of the five main brain regions involved in saccade generation: reticular formation saccadic burst generators, superior colliculus, cerebellum, basal ganglia and premotor cortical areas. We present the various topics these models are concerned with: location of the feedback loop, multimodal saccades, long-term adaptation, on the fly trajectory correction, strategy and metrics selection, short-term spatial memory, transformations between retinocentric and craniocentric reference frames, sequence learning, to name the principle ones. Our objective is to provide a global view of the whole system. Indeed, narrowing too much the modelled areas while trying to explain too much data is a recurrent problem that should be avoided. Moreover, beyond the multiple research topics remaining to be solved locally, questions regarding the operation of the whole structure can now be addressed by building on the existing models.

Feed-forward associative learning for volitional movement control

One of the most difficult problems in motor learning is determining the source of a learning signal, sometimes called an error signal. This problem is hidden in the adaptations of simple reflexive movements by attributing its source to sensory organs. The feed-forward associative motor learning theory proposed here attributes the source to the movement system itself. When a subject performs a corrective movement after his primary movement, the proposed neural learning device learns to associate the primary motor command with the corrective motor command by using a place-coding system. In the subsequent trials, the primary movement will involve a correction due to the participation of this mechanism, thus resulting in better performance. The theory assumes three conditions, namely, that a motor center and the learning device share the same place-encoded motor information; the motor center issues a command and a learning signal simultaneously from the same unit; and a learning signal issued with a corrective command has a heterosynaptic interaction with the previous primary command. The cerebellum is a reasonable candidate for the device satisfying these conditions. The reaction time of a corrective movement, usually 100-300 ms, almost satisfies the coincidence condition for long-term depression of the granule-to-Purkinje synapses. As an application, this theory is demonstrated to account for behavioral results regarding saccadic adaptation.

Discharge of monkey nucleus reticularis tegmenti pontis neurons changes during saccade adaptation

Saccade accuracy is maintained by adaptive mechanisms that continually modify saccade amplitude to reduce dysmetria. Previous studies suggest that adaptation occurs upstream of the caudal fastigial nucleus (CFN), the output of the oculomotor cerebellar vermis but downstream from the superior colliculus (SC). The nucleus reticularis tegmenti pontis (NRTP) is a major source of afferents to both the oculomotor vermis and the CFN and in turn receives direct input from the SC. Here we examine the activity of NRTP neurons in four rhesus monkeys during behaviorally induced changes in saccade amplitude to assess whether their discharge might reveal adaptation mechanisms that mediate changes in saccade amplitude. During amplitude decrease adaptation (average, 22%), the gradual reduction of saccade amplitude was accompanied by an increase in the number of spikes in the burst of 19/34 neurons (56%) and no change for 15 neurons (44%). For the neurons that increased their discharge, the additional spikes were added at the beginning of the saccadic burst and adaptation also delayed the peak-firing rate in some neurons. Moreover, after amplitude reduction, the movement fields changed shape in all 15 open field neurons tested. Our data show that saccadic amplitude reduction affects the number of spikes in the burst of more than half of NRTP neurons tested, primarily by increasing burst duration not frequency. Therefore adaptive changes in saccade amplitude are reflected already at a major input to the oculomotor cerebellum.

Computation of inverse functions in a model of cerebellar and reflex pathways allows to control a mobile mechanical segment

The command and control of limb movements by the cerebellar and reflex pathways are modeled by means of a circuit whose structure is deduced from functional constraints. One constraint is that fast limb movements must be accurate although they cannot be continuously controlled in closed loop by use of sensory signals. Thus, the pathways which process the motor orders must contain approximate inverse functions of the bio-mechanical functions of the limb and of the muscles. This can be achieved by means of parallel feedback loops, whose pattern turns out to be comparable to the anatomy of the cerebellar pathways. They contain neural networks able to anticipate the motor consequences of the motor orders, modeled by artificial neural networks whose connectivity is similar to that of the cerebellar cortex. These networks learn the direct biomechanical functions of the limbs and muscles by means of a supervised learning process. Teaching signals calculated from motor errors are sent to the learning sites, as, in the cerebellum, complex spikes issued from the inferior olive are conveyed to the Purkinje cells by climbing fibers. Learning rules are deduced by a differential calculation, as classical gradient rules, and they account for the long term depression which takes place in the dendritic arborizations of the Purkinje cells. Another constraint is that reflexes must not impede voluntary movements while remaining at any instant ready to oppose perturbations. Therefore, efferent copies of the motor orders are sent to the interneurones of the reflexes, where they cancel the sensory-motor consequences of the voluntary movements. After learning, the model is able to drive accurately, both in velocity and position, angular movements of a rod actuated by two pneumatic McKibben muscles. Reflexes comparable to the myotatic and tendinous reflexes, and stabilizing reactions comparable to the cerebellar sensory-motor reactions, reduce efficiently the effects of perturbing torques. These results allow to link the behavioral concepts of the equilibrium-point "lambda model" [J Motor Behav 18 (1986) 17] with anatomical and physiological features: gains of reflexes and sensori-motor reactions set the slope of the "invariant characteristic," and efferent copies set the "threshold of the stretch reflex." Thus, mathematical and physical laws account for the raison d'etre of the inhibitory nature of Purkinje cells and for the conspicuous anatomical pattern of the cerebellar pathways. These properties of these pathways allow to perform approximate inverse calculations after learning of direct functions, and insure also the coordination of voluntary and reflex motor orders.

The characteristics and neuronal substrate of saccadic eye movement plasticity

Saccadic eye movements are shifts in the direction of gaze that rapidly and accurately aim the fovea at targets of interest. Saccades are so brief that visual feedback cannot guide them to their targets. Therefore, the saccadic motor command must be accurately specified in advance of the movement and continually modified to compensate for growth, injury, and aging, which otherwise would produce dysmetric saccades. When a persistent dysmetria occurs in subjects with muscle weakness or neural damage or is induced in normal primates by the surreptitious jumping of a target forward or backward as a saccade is made to acquire the target, saccadic amplitude changes to reduce the dysmetria. Adaptation of saccadic amplitude or direction occurs gradually and is retained in the dark, thus representing true motor plasticity. Saccadic adaptation is more rapid in humans than in monkeys, usually is incomplete in both species, and is slower and less robust for amplitude increases than decreases. Adaptation appears to be motor rather than sensory. In humans, adaptation of saccades that would seem to require more sensory-motor processing does not transfer to saccades that seem to require less, suggesting the existence of distributed adaptation loci. In monkeys, however, transfer from more simple to more complex saccades is robust, suggesting a common adaptation site. Neurophysiological data from both species indicate that the oculomotor cerebellum is crucial for saccadic adaptation. This review shows that the precise, voluntary behaviors known as saccadic eye movements provide an alternative to simple reflexes for the study of the neuronal basis of motor learning.

Cerebellar learning of bio-mechanical functions of extra-ocular muscles: modeling by artificial neural networks

A control circuit is proposed to model the command of saccadic eye movements. Its wiring is deduced from a mathematical constraint, i.e. the necessity, for motor orders processing, to compute an approximate inverse function of the bio-mechanical function of the moving plant, here the bio-mechanics of the eye. This wiring is comparable to the anatomy of the cerebellar pathways. A predicting element, necessary for inversion and thus for movement accuracy, is modeled by an artificial neural network whose structure, deduced from physical constraints expressing the mechanics of the eye, is similar to the cell connectivity of the cerebellar cortex. Its functioning is set by supervised reinforcement learning, according to learning rules aimed at reducing the errors of pointing, and deduced from a differential calculation. After each movement, a teaching signal encoding the pointing error is distributed to various learning sites, as is, in the cerebellum, the signal issued from the inferior olive and conveyed to various cell types by the climbing fibers. Results of simulations lead to predict the existence of a learning site in the glomeruli. After learning, the model is able to accurately simulate saccadic eye movements. It accounts for the function of the cerebellar pathways and for the final integrator of the oculomotor system. The novelty of this model of movement control is that its structure is entirely deduced from mathematical and physical constraints, and is consistent with general anatomy, cell connectivity and functioning of the cerebellar pathways. Even the learning rules can be deduced from calculation, and they reproduce long term depression, the learning process which takes place in the dendritic arborization of the Purkinje cells. This approach, based on the laws of mathematics and physics, appears thus as an efficient way of understanding signal processing in the motor system.

A neural model of saccadic eye movement control explains task-specific adaptation

Multiple brain learning sites are needed to calibrate the accuracy of saccadic eye movements. This is true because saccades can be made reactively to visual cues, attentively to visual or auditory cues, or planned in response to memory cues using visual, parietal, and prefrontal cortex, as well as superior colliculus, cerebellum, and reticular formation. The organization of these sites can be probed by displacing a visual target during a saccade. The resulting adaptation typically shows incomplete and asymmetric transfer between different tasks. A neural model of saccadic system learning is developed to explain these data, as well as data about saccadic coordinate changes.

Model of the control of saccades by superior colliculus and cerebellum

Experimental evidence indicates that the superior colliculus (SC) is important but neither necessary nor sufficient to produce accurate saccadic eye movements. Furthermore both clinical and experimental evidence points to the cerebellum as an indispensable component of the saccadic system. Accordingly, we have devised a new model of the saccadic system in which the characteristics of saccades are determined by the cooperation of two pathways, one through the SC and the other through the cerebellum. Both pathways are influenced by feedback information: the feedback determines the decay of activity for collicular neurons and the timing of the activation for cerebellar neurons. We have modeled three types of cells (burst, buildup, and fixation neurons) found in the intermediate layers of the superior colliculus. We propose that, from the point of view of motor execution, the burst neurons and the buildup neurons are not functionally distinct with both providing a directional drive to the brain stem circuitry. The fixation neurons determine the onset of the saccade by disfacilitating the omnipause neurons in the brain stem. Excluding noise-related variations, the ratio of the horizontal to the vertical components of the collicular drive is fixed throughout the saccade (i.e., its direction is fixed); the duration of the drive is such that it always would produce hypermetric movements. The cerebellum plays three roles: first, it provides an additional directional drive, which improves the acceleration of the eyes; second, it keeps track of the progress of the saccade toward the target; and third, it ends the saccade by choking off the collicular drive. The drive provided by the cerebellum can be adjusted in direction to exert a directional control over the saccadic trajectory. We propose here a control mechanism that incorporates a spatial displacement integrator in the cerebellum; under such conditions, we show that a partial directional control arises automatically. Our scheme preserves the advantages of several previous models of the saccadic system (e.g., the lack of a spatial-to-temporal transformation between the SC and the brain stem; the use of efference copy feedback to control the saccade), without incurring many of their drawbacks, and it accounts for a large amount of experimental data.

Effects of lesions of the oculomotor vermis on eye movements in primate: saccades

We studied the effects on saccades of ablation of the dorsal cerebellar vermis (lesions centered on lobules VI and VII) in three monkeys in which the deep cerebellar nuclei were spared. One animal, with a symmetrical lesion, showed bilateral hypometric horizontal saccades. Two animals, with asymmetrical lesions, showed hypometric ipsilateral saccades, and saccades to vertically positioned targets were misdirected, usually deviating away from the side to which horizontal saccades were hypometric. Postlesion, all animals showed an increase (2- to 5-fold) in trial-to-trial variability of saccade amplitude. They also showed a change in the ratio of the amplitudes of centripetal to centrifugal saccades (orbital-position effect); usually centrifugal saccades became smaller. In the two animals with asymmetrical lesions, for saccades in the hypometric direction, latencies were markedly increased (up to approximately 500 ms). There was also an absence of express and anticipatory saccades in the hypometric direction. When overall saccade latency was increased, centrifugal saccades became relatively more delayed than centripetal saccades. The dynamic characteristics of saccades were affected to some extent in all monkeys with changes in peak velocity, eye acceleration, and especially eye deceleration. There was relatively little effect of orbital position on saccade dynamics, however, with the exception of one animal that showed an orbital position effect for eye acceleration. In a double-step adaptation paradigm, animals showed an impaired ability to adaptively adjust saccade amplitude, though increased amplitude variability postlesion may have played a role in this deficit. During a single training session, however, the latency to corrective saccades-which had been increased postlesion-gradually decreased and so enabled the animal to reach the final position of the target more quickly. Overall, both in the early postlesion period and during recovery, changes in saccade amplitude and latency tended to vary together but not with changes in saccade dynamics or adaptive capability, both of which behaved relatively independently. These findings suggest that the cerebellum can adjust saccade amplitude and saccade dynamics independently. Our results implicate the cerebellar vermis directly in every aspect of the on-line control of saccades: initiation (latency), accuracy (amplitude and direction), and dynamics (velocity and acceleration) and also in the acquisition of adaptive ocular motor behavior.

Effects of three stimulus parameters on eye position in cerebral palsied adults

Bilateral eye position was measured in 6 cerebral palsied adults to assess the effects of stimulus dimensions (horizontal, vertical), amplitude (+/- 4 degrees, +/- 6 degrees, +/- 8 degrees), and frequency (0.3, 0.5, 0.7 Hz) on saccadic and pursuit movements. The head-free, corneal reflection method was used for 54 10-sec. trials of square, triangle, and sine wave stimuli. Shared variance between each eye's position and the stimulus was tested by Wilcoxon T (dimension) and Friedman analysis of variance (amplitude, frequency) showing that the effects of saccadic and pursuit dimension and amplitude were individualized with regard to subject and right and left eye positions. The bilateral eye position of 5 of 6 subjects was affected by saccadic frequency; pursuit frequency affected bilateral eye position of 4 of 6 subjects. The lowest shared variance (critical difference in ranks) was at 0.7 Hz. The results are discussed with regard to subjects' disability, stimulus velocity, and frequency of directional reversal. Reversal may be the most critical stimulus property.

A neural model of multimodal adaptive saccadic eye movement control by superior colliculus

How does the saccadic movement system select a target when visual, auditory, and planned movement commands differ? How do retinal, head-centered, and motor error coordinates interact during the selection process? Recent data on superior colliculus (SC) reveal a spreading wave of activation across buildup cells the peak activity of which covaries with the current gaze error. In contrast, the locus of peak activity remains constant at burst cells, whereas their activity level decays with residual gaze error. A neural model answers these questions and simulates burst and buildup responses in visual, overlap, memory, and gap tasks. The model also simulates data on multimodal enhancement and suppression of activity in the deeper SC layers and suggests a functional role for NMDA receptors in this region. In particular, the model suggests how auditory and planned saccadic target positions become aligned and compete with visually reactive target positions to select a movement command. For this to occur, a transformation between auditory and planned head-centered representations and a retinotopic target representation is learned. Burst cells in the model generate teaching signals to the spreading wave layer. Spreading waves are produced by corollary discharges that render planned and visually reactive targets dimensionally consistent and enable them to compete for attention to generate a movement command in motor error coordinates. The attentional selection process also helps to stabilize the map-learning process. The model functionally interprets cells in the superior colliculus, frontal eye field, parietal cortex, mesencephalic reticular formation, paramedian pontine reticular formation, and substantia nigra pars reticulata.

Specificity of saccadic adaptation in three-dimensional space

The saccadic system is known to exhibit a considerable degree of short-term plasticity. Earlier studies have shown that saccadic adaptation, rather than being a global process affecting all saccades equally, has a certain degree of spatial resolution. Its localized nature has become apparent from studies in the frontal plane which have shown that short-term saccadic adaptation, induced along a given meridian, transfers to only a limited range of neighbouring directions. Considering that most natural gaze shifts also have a depth component, we investigated whether the directional specificity of the saccadic adaptive system can be generalized to three-dimensional (3-D) space. Binocular eye movements were recorded in seven subjects while they made saccades to visual stimuli in the horizontal plane of regard. Experiments began by recording baseline saccades, all starting from the same fixation point to either a farther target (far saccades) or an equally eccentric nearer target (near saccades). Next, by displacing the target intra-saccadically in opposite directions in alternating far and near trials, we attempted to simultaneously reduce the gain of the far saccades while increasing the gain of the near saccades. These experiments, aimed at eliciting a state of differential gain, were specifically designed to adapt only the saccadic response, since targets were shifted along corresponding iso-vergence circles. To investigate the effect of varying the radial direction difference, similar differential gain adaptation experiments were conducted in the frontal plane for saccades along two different meridians. Our results show that when the saccadic system is pressured, it is capable of adopting different gains simultaneously for equal-direction saccades to different depth planes. Similarly, opposite gain adaptation can also be achieved in the frontal plane, but only if radial saccade directions are sufficiently separated. The fact that short-term saccadic adaptation can be shown to be directionally specific in two perpendicular planes suggests that the adaptation process is restricted to a limited volume of 3-D oculomotor space.

Prediction of complex two-dimensional trajectories by a cerebellar model of smooth pursuit eye movement

A neural network model based on the anatomy and physiology of the cerebellum is presented that can generate both simple and complex predictive pursuit, while also responding in a feedback mode to visual perturbations from an ongoing trajectory. The model allows the prediction of complex movements by adding two features that are not present in other pursuit models: an array of inputs distributed over a range of physiologically justified delays, and a novel, biologically plausible learning rule that generated changes in synaptic strengths in response to retinal slip errors that arrive after long delays. To directly test the model, its output was compared with the behavior of monkeys tracking the same trajectories. There was a close correspondence between model and monkey performance. Complex target trajectories were created by summing two or three sinusoidal components of different frequencies along horizontal and/or vertical axes. Both the model and the monkeys were able to track these complex sum-of-sines trajectories with small phase delays that averaged 8 and 20 ms in magnitude, respectively. Both the model and the monkeys showed a consistent relationship between the high- and low-frequency components of pursuit: high-frequency components were tracked with small phase lags, whereas low-frequency components were tracked with phase leads. The model was also trained to track targets moving along a circular trajectory with infrequent right-angle perturbations that moved the target along a circle meridian. Before the perturbation, the model tracked the target with very small phase differences that averaged 5 ms. After the perturbation, the model overshot the target while continuing along the expected nonperturbed circular trajectory for 80 ms, before it moved toward the new perturbed trajectory. Monkeys showed similar behaviors with an average phase difference of 3 ms during circular pursuit, followed by a perturbation response after 90 ms. In both cases, the delays required to process visual information were much longer than delays associated with nonperturbed circular and sum-of-sines pursuit. This suggests that both the model and the eye make short-term predictions about future events to compensate for visual feedback delays in receiving information about the direction of a target moving along a changing trajectory. In addition, both the eye and the model can adjust to abrupt changes in target direction on the basis of visual feedback, but do so after significant processing delays.

Monkey superior colliculus activity during short-term saccadic adaptation

This article concerns the neural mechanisms that underlie short-term saccadic adaptation in the rhesus monkey. By means of a consistent intrasaccadic target displacement, the relation between visual input and motor output was gradually changed in three monkeys, such that they made hypometric saccades. During this process, the activity of saccade-related burst neurons in the intermediate and deep layers of the Superior Colliculus (SC) was recorded in two of the monkeys. Our findings show that, like in humans, only saccades evoked within a restricted field around the adaptation target were adapted. However, unlike in humans, the kinematic properties of adapted saccades also changed systematically during the adaptation process. Typically, adapted saccades were slower and had a longer duration than would be expected on the basis of the main sequence for nonadapted visually guided movements. During adaptation, saccade-related activity of units in the SC remained appropriate for the saccade that was required to foveate the initial target, rather than for the saccade that was actually made. This means that adaptation caused a dissociation between SC activity and the ensuing saccade. Thus, the activity of the colliculus was better described in "required eye displacement coordinates" than in "actual eye displacement coordinates." Our data provide further evidence for the hypothesis that short-term saccadic adaptation acts at a level downstream from the SC, presumably at a stage that determines the kinematics of saccadic eye movements.

Effects of midline and lateral cerebellar lesions on motor coordination and spatial orientation

Rats were lesioned in the midline cerebellum, comprising the vermis and fastigial nucleus, or the lateral cerebellum, comprising the cerebellar hemispheres and dentate nucleus, and evaluated in a series of motor and non-motor learning tests. Rats with midline lesions had difficulty in maintaining their equilibrium on a bridge and were slower before turning upward and traversed less squares on an inclined grid. They were not impaired for muscle strength when suspended from a horizontal wire. Rats with lateral lesions had milder deficits on the bridge and were not affected in the other two tests. In the Morris water maze test, rats with lateral lesions were deficient in spatial orientation, whereas rats with midline lesions were deficient in visuomotor coordination. Lateral lesions had no effects on visual discrimination learning. These results illustrate the differential influence of midline as opposed to lateral cerebellar regions on both motor and non-motor behaviors. Fastigial nucleus lesions decreased the time spent in equilibrium and latencies before falling on the bridge and the distance travelled along the inclined grid but had no effect on muscle strength when suspended from the horizontal string. Quadrant entries and escape latencies were higher in rats with fastigial lesions during the hidden platform condition of the Morris water maze but not during the visible platform condition. It is concluded that fastigial-lesioned rats are impaired in equilibrium and spatial orientation but with repeated trials learn to improve their performances.