Vestibular processing during natural self-motion: implications for perception and action

Abstract

How the brain computes accurate estimates of our self-motion relative to the world and our orientation relative to gravity in order to ensure accurate perception and motor control is a fundamental neuroscientific question. Recent experiments have revealed that the vestibular system encodes this information during everyday activities using pathway-specific neural representations. Furthermore, new findings have established that vestibular signals are selectively combined with extravestibular information at the earliest stages of central vestibular processing in a manner that depends on the current behavioural goal. These findings have important implications for our understanding of the brain mechanisms that ensure accurate perception and behaviour during everyday activities and for our understanding of disorders of vestibular processing.

Figure 1.. Overview of the vestibular labyrinth and central pathways.

a| The vestibular labyrinth comprises five end organs: the three semicircular canals and the two otoliths (utricle and the saccule). In mammals, there are two types of hair cells within each of the vestibular sensory organs: cylindrical type II hair cells and the phylogenetically older flask-shaped type I hair cells. Peripheral afferents in the VIII cranial nerve innervate hair cells and carry head movement signals to the vestibular nuclei and to some regions of the vestibular cerebellum. Each semicircular canal afferent innervates one of the three canals and encodes information about angular head motion. Otolith afferents innervate either the utricle or saccule and encode information about translational acceleration. Notably, otolith afferents respond to the inertial forces produced by translational motion through the environment or by changes in head orientation relative to gravity. Both canal and otolith afferent fibers are classified on the basis of the regularity of their resting discharge (reviewed in refs ^,). In general, irregular afferents have larger axons and preferentially transmit information from either the type I hair cells located at the center of neuroepithelium (known as C-irregulars) or from both type I hair cells and type II hair cells (known as dimorphic or D-irregulars), whereas regular afferents preferentially provide bouton (B) endings to type II hair cells. b| The vestibular system makes essential contributions to our perception of self-motion and ability to navigate, as well as to vital reflex pathways (the vestibulo-ocular reflex (VOR) and the vestibulo-spinal reflex (VSR)). Vestibular information is sent to cortex via two ascending vestibular thalamocortical pathways: the anterior vestibulothalamic pathway, comprised of projections from the vestibular nuclei (VN) to the nucleus prepositus and supragenual nucleus (NPH/SGN) and then on to the anterior dorsal thalamus (ADN) via the head direction (HD) network and the posterior vestibulothalamic pathway comprised of projections from the vestibular nuclei through the ventral posterior lateral nucleus (VPL).

Figure 2.. Motor and multisensory integration in the vestibular nuclei.

Whereas vestibular afferents encode head movements regardless of the behavioral goal, the responses of central neurons mediating the vestibulo-ocular reflex (VOR) and of those projecting to vestibulo-spinal (ascending) pathways are differentially modulated by extravestibular information. a| The schematic illustrates the pathway that mediates the VOR reflex, which acts to stablize gaze in response to head motion. When the goal of the head movement is to voluntarily redirect gaze, the responses of central VOR neurons are suppressed. Specifically, the responses of the position-vestibular-pause (PVP) neurons that mediate the intermediate link in the VOR pathway are strongly inhibited by input from the premotor saccadic pathway that drives gaze shifts via its direct projections to extraocular motoneurons. This inhibition is denoted on the schematic as a ‘gate’ that closes when the goal is to voluntarily redirect (rather than stabilize) gaze. Inset illustrates the reduced the firing rate of a PVP neuron during a voluntary gaze shift relative to a firing rate prediction based on its sensitivity to the same type of head motion during the VOR The inhibitory saccadic drive can be accounted for by known brainstem mechanisms. Specifically, burst neurons in the brainstem paramedian pontine reticular formation generate a burst of spikes to drive horizontal saccadic eye movements, and send direct inhibitory projections to the vestibular nuclei^{, ,} b| The schematic illustrates the vestibulo-spinal pathways that mediate postural stabilization through the vestibulo-spinal reflex (VSR), self-motion perception and navigation. Within this pathway, vestibular-only (VO) neurons respond to passive head motion. However, when the behavioral goal is to generate active head motion, the responses of central VO neurons are suppressed. Specifically, VO neurons receive a strong inhibitory cancellation signal when there is a match between the expected sensory consequence of the neck motor command and the actual neck proprioceptive feedback. Such a match functions to cancel vestibular reafference and suppress VO neuron responses ^,,. Inset illustrates the firing rate response of a VO neuron during combined active and passive head motion. In this condition, monkeys generated active head-on-body movements (red arrow) while being passively rotated by the vestibular turntable (blue arrow), such that head motion is the sum of the passive sinusoidal stimulus and the monkey’s active head movement. The inset illustrates the selective cancellation of the VO neuron’s response to the active component of the head motion. Specifically, the neuron responds to the passive sinusoidal component of head motion stimulus (superimposed blue trace), but is unresponsive to the active component of head motion. The dashed red trace shows the firing rate prediction based on total head motion. c| Schematic illustrating the generation of the reafference cancellation signal that suppresses VO neuron activity in the cerebellum. During active movement, the brain computes an internal (forward) model of the expected sensory consequences of a motor command. This estimate is compared with the actual sensory inflow to compute the sensory prediction error (SPE). When there is match between expected and actual sensory inflow (that is, SPE= 0), vestibular reafference is suppressed^,,. Inset in part a is adapted, with permission from REF . Inset in part b is adapted, with permission from REF .

Figure 3.. Internal models of self-motion in the vestibular cerebellum

a| Schematic demonstrating stimulation of the vestibular (head-centered) and proprioceptive (body-centered) systems in a monkey. The vestibular system alone is stimulated by passively rotating the head and body together relative to space (whole body rotation, top panel) and the proprioceptive system alone is stimulated by passively rotating the body under the head, which remains stationary (middle panel). By contrast, passive rotation of the head relative to the body (bottom panel) produces combined stimulation of the proprioceptive and vestibular systems, thereby allowing us to investigate the transformation of vestibular input from head- to body-centered coordinates. b| The responses of an example deep cerebellar nuclei neuron during passive and active (voluntary) motion paradigms. The top traces illustrate head velocity, the bottom traces show neuronal firing rate responses and the dashed red line indicates a prediction of the firing rate in the active condition based on the neuron’s sensitivity to passive motion. The neurons show robust modulation by passive motion, but their responses are minimal when the same motion is actively-generated. In the lower panels, grey shading corresponds to the average firing rates and standard deviations for the same 10 movements. Overlaying blue and red lines show the average firing rate responses to passive versus active motion, respectively. c| When the relationship between the head motor command and resultant movement is altered by applying a resistive load (torque), neuronal vestibular sensitivities to head motion during active head movements initially increase to levels measured during passive head motion. They then gradually decrease to those measured during active head motion before torque application. Once the brain’s internal model has been updated to accommodate the new relationship between the voluntary head motor command and resultant movement, neuronal sensitivities during active trials in which the load is removed (catch trials) are comparable to the neuronal sensitivity during passive head movements. Notably, the re-emergence of afferent suppression during learning — represented by a decrease of the normalized sensitivity of neuronal responses — follows the same time course as the corresponding change in head movement error (not shown). Part b is adapted, with permission from REF . Part c is adapted, with permission from REF .

Figure 4.. Vestibular processing for self-motion perception

a| A comparison of the average behavioral and neuronal thresholds for the detection of whole-body motion. The results of psychophysical studies in humans in which subjects experienced applied whole body rotation in darkness around an earth vertical axis (as illustrated to the right) at different frequencies are shown. Superimposed for comparison are the neuronal detection thresholds (+/− SEM) measured in alert monkeys for populations of regular and irregular semicircular canal afferents as well as central vestibular-only (VO) neurons. Also shown is the neuronal detection threshold computed for a pooled population of 12 VO neurons, assuming independent noise. Notably, in order to recover the level of information required to explain psychophysical performance, ascending vestibular pathways must integrate information from large populations of vestibular nuclei neurons. Psychophysical and neuronal data from REF and REF respectively . b| A schematic drawing illustrating the regions of cortex that receive inputs from vestibular nuclei (left panel) and those that project back to the vestibular nuclei (right panel). Striped regions denote areas that areas receive direct input from the vestibular nuclei. c| Comparison of neuronal responses to active rotations and translations (normalized relative to responses to comparable passive motion, dashed line) at four successive stages of vestibular processing: the vestibular afferents (data from REFs ^,), the vestibular nuclei (data from REFs ^,), the fastigial nucleus in the cerebellum (data from REF) and the thalamus (data from REF ). Note that neurons at each subsequent stage of central vestibular processing are increasingly selective to passive self-motion (vestibular exafference). Error bars show SEM. FEF, frontal eye field; MST, medial superior temporal area; PIVC, parietoinsular vestibular cortex; VIP, ventral intraparietal area. Part a is adapted, with permission from REF . Part b is adapted, with permission from REF . Part c is adapted, with permission from REF .