Neural network simulations of the primate oculomotor system. V. Eye-head gaze shifts

Pathways for Naturalistic Looking Behavior in Primate I: Behavioral Characteristics and Brainstem Circuits

Organisms control their visual worlds by moving their eyes, heads, and bodies. This control of "gaze" or "looking" is key to survival and intelligence, but our investigation of the underlying neural mechanisms in natural conditions is hindered by technical limitations. Recent advances have enabled measurement of both brain and behavior in freely moving animals in complex environments, expanding on historical head-fixed laboratory investigations. We juxtapose looking behavior as traditionally measured in the laboratory against looking behavior in naturalistic conditions, finding that behavior changes when animals are free to move or when stimuli have depth or sound. We specifically focus on the brainstem circuits driving gaze shifts and gaze stabilization. The overarching goal of this review is to reconcile historical understanding of the differential neural circuits for different "classes" of gaze shift with two inconvenient truths. (1) "classes" of gaze behavior are artificial. (2) The neural circuits historically identified to control each "class" of behavior do not operate in isolation during natural behavior. Instead, multiple pathways combine adaptively and non-linearly depending on individual experience. While the neural circuits for reflexive and voluntary gaze behaviors traverse somewhat independent brainstem and spinal cord circuits, both can be modulated by feedback, meaning that most gaze behaviors are learned rather than hardcoded. Despite this flexibility, there are broadly enumerable neural pathways commonly adopted among primate gaze systems. Parallel pathways which carry simultaneous evolutionary and homeostatic drives converge in superior colliculus, a layered midbrain structure which integrates and relays these volitional signals to brainstem gaze-control circuits.

Neural encoding of instantaneous kinematics of eye-head gaze shifts in monkey superior Colliculus

The midbrain superior colliculus is a crucial sensorimotor stage for programming and generating saccadic eye-head gaze shifts. Although it is well established that superior colliculus cells encode a neural command that specifies the amplitude and direction of the upcoming gaze-shift vector, there is controversy about the role of the firing-rate dynamics of these neurons during saccades. In our earlier work, we proposed a simple quantitative model that explains how the recruited superior colliculus population may specify the detailed kinematics (trajectories and velocity profiles) of head-restrained saccadic eye movements. We here show that the same principles may apply to a wide range of saccadic eye-head gaze shifts with strongly varying kinematics, despite the substantial nonlinearities and redundancy in programming and execute rapid goal-directed eye-head gaze shifts to peripheral targets. Our findings could provide additional evidence for an important role of the superior colliculus in the optimal control of saccades.

Dynamic control of eye-head gaze shifts by a spiking neural network model of the superior colliculus

INTRODUCTION

To reorient gaze (the eye's direction in space) towards a target is an overdetermined problem, as infinitely many combinations of eye- and head movements can specify the same gaze-displacement vector. Yet, behavioral measurements show that the primate gaze-control system selects a specific contribution of eye- and head movements to the saccade, which depends on the initial eye-in-head orientation. Single-unit recordings in the primate superior colliculus (SC) during head-unrestrained gaze shifts have further suggested that cells may encode the instantaneous trajectory of a desired straight gaze path in a feedforward way by the total cumulative number of spikes in the neural population, and that the instantaneous gaze kinematics are thus determined by the neural firing rates. The recordings also indicated that the latter is modulated by the initial eye position. We recently proposed a conceptual model that accounts for many of the observed properties of eye-head gaze shifts and on the potential role of the SC in gaze control.

METHODS

Here, we extend and test the model by incorporating a spiking neural network of the SC motor map, the output of which drives the eye-head motor control circuitry by linear cumulative summation of individual spike effects of each recruited SC neuron. We propose a simple neural mechanism on SC cells that explains the modulatory influence of feedback from an initial eye-in-head position signal on their spiking activity. The same signal also determines the onset delay of the head movement with respect to the eye. Moreover, the downstream eye- and head burst generators were taken to be linear, as our earlier work had indicated that much of the non-linear main-sequence kinematics of saccadic eye movements may be due to neural encoding at the collicular level, rather than at the brainstem.

RESULTS AND DISCUSSION

We investigate how the spiking activity of the SC population drives gaze to the intended target location within a dynamic local gaze-velocity feedback circuit that yields realistic eye- and head-movement kinematics and dynamic SC gaze-movement fields.

Maps and sensorimotor transformations for eye-head gaze shifts: Role of the midbrain superior colliculus

Single-unit recordings in head-restrained monkeys indicated that the population of saccade-related cells in the midbrain Superior Colliculus (SC) encodes the kinematics of desired straight saccade trajectories by the cumulative number of spikes. In addition, the nonlinear main sequence of saccades (their amplitude-peak velocity saturation) emerges from a spatial gradient of peak-firing rates of collicular neurons, rather than from neural saturation at brainstem burst generators. We here extend this idea to eye-head gaze shifts and illustrate how the cumulative spike-count in head-unrestrained monkeys relates to the desired gaze trajectory and its kinematics. We argue that the output of the motor SC is an abstract desired gaze-motor signal, which drives in a feedforward way the instantaneous kinematics of ongoing gaze shifts, including the strong influence of initial eye position on gaze kinematics. We propose that the neural population acts as a vectorial gaze pulse-generator for eye-head saccades, which is subsequently decomposed into signals that drive both motor systems in appropriate craniocentric reference frames within a dynamic gaze-velocity feedback loop.

Response Properties of Motor Equivalence Neurons of the Primate Premotor Cortex

To study the response properties of cells that could participate in eye-hand coordination we trained two macaque monkeys to perform center-out saccades and pointing movements with their right or left forelimb toward visual targets presented on a video display. We analyzed the phasic movement related discharges of neurons of the periarcuate cortex that fire before and during saccades and movements of the hand whether accompanied by movements of the other effector or not. Because such cells could encode an abstract form of the desired displacement vector without regard to the effector that would execute the movement we refer to such cells as motor equivalence neurons (Meq). Most of them (75%) were found in or near the smooth pursuit region and the grasp related region in the caudal bank of the arcuate sulcus. The onset of their phasic discharges preceded saccades by about 70 ms and hand movements by about 150 ms and was often correlated to both the onset of saccades and the onset of hand movements. The on-direction of Meq cells was uniformly distributed without preference for ipsiversive or contraversive movements. In about half of the Meq cells the preferred direction for saccades was the preferred direction for hand movements as well. In the remaining cells the difference was considerable (>90 deg), and the on-direction for eye-hand movements resembled that for isolated saccades in some cells and for isolated hand movements in others. A three layer neural network model that used Meq cells as its input layer showed that the combination of effector invariant discharges with non-invariant discharges could help reduce the number of decoding errors when the network attempts to compute the correct movement metrics of the right effector.

Modeling eye-head gaze shifts in multiple contexts without motor planning

During gaze shifts, the eyes and head collaborate to rapidly capture a target (saccade) and fixate it. Accordingly, models of gaze shift control should embed both saccadic and fixation modes and a mechanism for switching between them. We demonstrate a model in which the eye and head platforms are driven by a shared gaze error signal. To limit the number of free parameters, we implement a model reduction approach in which steady-state cerebellar effects at each of their projection sites are lumped with the parameter of that site. The model topology is consistent with anatomy and neurophysiology, and can replicate eye-head responses observed in multiple experimental contexts: 1) observed gaze characteristics across species and subjects can emerge from this structure with minor parametric changes; 2) gaze can move to a goal while in the fixation mode; 3) ocular compensation for head perturbations during saccades could rely on vestibular-only cells in the vestibular nuclei with postulated projections to burst neurons; 4) two nonlinearities suffice, i.e., the experimentally-determined mapping of tectoreticular cells onto brain stem targets and the increased recruitment of the head for larger target eccentricities; 5) the effects of initial conditions on eye/head trajectories are due to neural circuit dynamics, not planning; and 6) "compensatory" ocular slow phases exist even after semicircular canal plugging, because of interconnections linking eye-head circuits. Our model structure also simulates classical vestibulo-ocular reflex and pursuit nystagmus, and provides novel neural circuit and behavioral predictions, notably that both eye-head coordination and segmental limb coordination are possible without trajectory planning.

Vestibulo-ocular reflex suppression during head-fixed saccades reveals gaze feedback control

Previous experiments have shown that the vestibulo-ocular reflex (VOR) is partially suppressed during large head-free gaze (gaze = eye-in-head + head-in-space) shifts when both the eyes and head are moving actively, on a fixed body, or when the eyes are moving actively and the head passively on a fixed body. We tested, in human subjects, the hypothesis that the VOR is also suppressed during gaze saccades made with en bloc, head and body together, rotations. Subjects made saccades by following a target light. During some trials, the chair rotated so as to move the entire body passively before, during, or after a saccade. The modulation of the VOR was a function of both saccade amplitude and the time of the head perturbation relative to saccade onset. Despite the perturbation, gaze remained accurate. Thus, VOR modulation is similar when gaze changes are programmed for the eyes alone or for the eyes and head moving together. We propose that the brain always programs a change in gaze using feedback based on gaze and head signals, rather than on separate eye and head trajectories.

Hierarchical control of two-dimensional gaze saccades

Coordinating the movements of different body parts is a challenging process for the central nervous system because of several problems. Four of these main difficulties are: first, moving one part can move others; second, the parts can have different dynamics; third, some parts can have different motor goals; and fourth, some parts may be perturbed by outside forces. Here, we propose a novel approach for the control of linked systems with feedback loops for each part. The proximal parts have separate goals, but critically the most distal part has only the common goal. We apply this new control policy to eye-head coordination in two-dimensions, specifically head-unrestrained gaze saccades. Paradoxically, the hierarchical structure has controllers for the gaze and the head, but not for the eye (the most distal part). Our simulations demonstrate that the proposed control structure reproduces much of the published empirical data about gaze movements, e.g., it compensates for perturbations, accurately reaches goals for gaze and head from arbitrary initial positions, simulates the nine relationships of the head-unrestrained main sequence, and reproduces observations from lesion and single-unit recording experiments. We conclude by showing how our model can be easily extended to control structures with more linked segments, such as the control of coordinated eye on head on trunk movements.

Human eye-head gaze shifts preserve their accuracy and spatiotemporal trajectory profiles despite long-duration torque perturbations that assist or oppose head motion

Humans routinely use coordinated eye-head gaze saccades to rapidly and accurately redirect the line of sight (Land MF. Vis Neurosci 26: 51–62, 2009). With a fixed body, the gaze control system combines visual, vestibular, and neck proprioceptive sensory information and coordinates two moving platforms, the eyes and head. Classic engineering tools have investigated the structure of motor systems by testing their ability to compensate for perturbations. When a reaching movement of the hand is subjected to an unexpected force field of random direction and strength, the trajectory is deviated and its final position is inaccurate. Here, we found that the gaze control system behaves differently. We perturbed horizontal gaze shifts with long-duration torques applied to the head that unpredictably either assisted or opposed head motion and very significantly altered the intended head trajectory. We found, as others have with brief head perturbations, that gaze accuracy was preserved. Unexpectedly, we found also that the eye compensated well—with saccadic and rollback movements—for long-duration head perturbations such that resulting gaze trajectories remained close to that when the head was not perturbed. However, the ocular compensation was best when torques assisted, compared with opposed, head motion. If the vestibuloocular reflex (VOR) is suppressed during gaze shifts, as currently thought, what caused invariant gaze trajectories and accuracy, early eye-direction reversals, and asymmetric compensations? We propose three mechanisms: a gaze feedback loop that generates a gaze-position error signal; a vestibular-to-oculomotor signal that dissociates self-generated from passively imposed head motion; and a saturation element that limits orbital eye excursion.

Learning the optimal control of coordinated eye and head movements

Various optimality principles have been proposed to explain the characteristics of coordinated eye and head movements during visual orienting behavior. At the same time, researchers have suggested several neural models to underly the generation of saccades, but these do not include online learning as a mechanism of optimization. Here, we suggest an open-loop neural controller with a local adaptation mechanism that minimizes a proposed cost function. Simulations show that the characteristics of coordinated eye and head movements generated by this model match the experimental data in many aspects, including the relationship between amplitude, duration and peak velocity in head-restrained and the relative contribution of eye and head to the total gaze shift in head-free conditions. Our model is a first step towards bringing together an optimality principle and an incremental local learning mechanism into a unified control scheme for coordinated eye and head movements.

Human beings and many other species redirect their gaze towards targets of interest through rapid gaze shifts known as saccades. These are made approximately three to four times every second, and larger saccades result from fast and concurrent movement of the animal's eyes and head. Experimental studies have revealed that during saccades, the motor system follows certain principles such as respecting a specific relationship between the relative contribution of eye and head motor systems to total gaze shift. Various researchers have hypothesized that these principles are implications of some optimality criteria in the brain, but it remains unclear how the brain can learn such an optimal behavior. We propose a new model that uses a plausible learning mechanism to satisfy an optimality criterion. We show that after learning, the model is able to reproduce motor behavior with biologically plausible properties. In addition, it predicts the nature of the learning signals. Further experimental research is necessary to test the validity of our model.

Electrical stimulation of the frontal eye fields in the head-free macaque evokes kinematically normal 3D gaze shifts

The frontal eye field (FEF) is a region of the primate prefrontal cortex that is central to eye-movement generation and target selection. It has been shown that neurons in this area encode commands for saccadic eye movements. Furthermore, it has been suggested that the FEF may be involved in the generation of gaze commands for the eye and the head. To test this suggestion, we systematically stimulated (with pulses of 300 Hz frequency, 200 ms duration, 30-100 μA intensity) the FEF of two macaques, with the head unrestrained, while recording three-dimensional (3D) eye and head rotations. In a total of 95 sites, the stimulation consistently elicited gaze-orienting movements ranging in amplitude from 2 to 172°, directed contralateral to the stimulation site, and with variable vertical components. These movements were typically a combination of eye-in-head saccades and head-in-space movements. We then performed a comparison between the stimulation-evoked movements and gaze shifts voluntarily made by the animal. The kinematics of the stimulation-evoked movements (i.e., their spatiotemporal properties, their velocity-amplitude relationships, and the relative contributions of the eye and the head as a function of movement amplitude) were very similar to those of natural gaze shifts. Moreover, they obeyed the same 3D constraints as the natural gaze shifts (i.e., modified Listing's law for eye-in-head movements). As in natural gaze shifts, saccade and vestibuloocular reflex torsion during stimulation-evoked movements were coordinated so that at the end of the head movement the eye-in-head ended up in Listing's plane. In summary, movements evoked by stimulation of the FEF closely resembled those of naturally occurring eye-head gaze shifts. Thus we conclude that the FEF explicitly encodes gaze commands and that the kinematic aspects of eye-head coordination are likely specified by downstream mechanisms.

Matching the oculomotor drive during head-restrained and head-unrestrained gaze shifts in monkey

High-frequency burst neurons in the pons provide the eye velocity command (equivalently, the primary oculomotor drive) to the abducens nucleus for generation of the horizontal component of both head-restrained (HR) and head-unrestrained (HU) gaze shifts. We sought to characterize how gaze and its eye-in-head component differ when an "identical" oculomotor drive is used to produce HR and HU movements. To address this objective, the activities of pontine burst neurons were recorded during horizontal HR and HU gaze shifts. The burst profile recorded on each HU trial was compared with the burst waveform of every HR trial obtained for the same neuron. The oculomotor drive was assumed to be comparable for the pair yielding the lowest root-mean-squared error. For matched pairs of HR and HU trials, the peak eye-in-head velocity was substantially smaller in the HU condition, and the reduction was usually greater than the peak head velocity of the HU trial. A time-varying attenuation index, defined as the difference in HR and HU eye velocity waveforms divided by head velocity [alpha = (H(hr) - E(hu))/H] was computed. The index was variable at the onset of the gaze shift, but it settled at values several times greater than 1. The index then decreased gradually during the movement and stabilized at 1 around the end of gaze shift. These results imply that substantial attenuation in eye velocity occurs, at least partially, downstream of the burst neurons. We speculate on the potential roles of burst-tonic neurons in the neural integrator and various cell types in the vestibular nuclei in mediating the attenuation in eye velocity in the presence of head movements.