Skip to main content
Journal of Neurophysiology logoLink to Journal of Neurophysiology
. 2014 May 21;112(4):870–889. doi: 10.1152/jn.00280.2014

Long-term deficits in motion detection thresholds and spike count variability after unilateral vestibular lesion

Xiong-Jie Yu 1,2,*, Jakob S Thomassen 1,2,*, J David Dickman 1,2, Shawn D Newlands 3, Dora E Angelaki 1,2,
PMCID: PMC4122753  PMID: 24848470

Abstract

The vestibular system operates in a push-pull fashion using signals from both labyrinths and an intricate bilateral organization. Unilateral vestibular lesions cause well-characterized motor deficits that are partially compensated over time and whose neural correlates have been traced in the mean response modulation of vestibular nuclei cells. Here we compare both response gains and neural detection thresholds of vestibular nuclei and semicircular canal afferent neurons in intact vs. unilateral-lesioned macaques using three-dimensional rotation and translation stimuli. We found increased stimulus-driven spike count variability and detection thresholds in semicircular canal afferents, although mean responses were unchanged, after contralateral labyrinth lesion. Analysis of trial-by-trial spike count correlations of a limited number of simultaneously recorded pairs of canal afferents suggests increased noise correlations after lesion. In addition, we also found persistent, chronic deficits in rotation detection thresholds of vestibular nuclei neurons, which were larger in the ipsilesional than the contralesional brain stem. These deficits, which persisted several months after lesion, were due to lower rotational response gains, whereas spike count variability was similar in intact and lesioned animals. In contrast to persistent deficits in rotation threshold, translation detection thresholds were not different from those in intact animals. These findings suggest that, after compensation, a single labyrinth is sufficient to recover motion sensitivity and normal thresholds for the otolith, but not the semicircular canal, system.

Keywords: vestibular nuclei, vestibular afferents, threshold, variability, interneuronal correlation


unilateral vestibular lesions cause well-characterized postural and motor deficits (reviewed in Curthoys 2000; Dieringer 1995; Vibert et al. 1997). Unilateral labyrinthectomy (LB), in particular, results in large “static” symptoms, including spontaneous nystagmus with slow phase toward the lesioned side and static head tilt, both of which recover within a few days, depending on the species (Darlington et al. 2002; de Waele et al. 1989; Newlands et al. 2005; Precht et al. 1966; Sadeghi et al. 2006; Sirkin et al. 1984; Smith and Curthoys 1989). Despite this initial fast recovery, substantial asymmetries in the high-frequency rotational vestibuloocular reflex (rVOR) persist for the long term and are more pronounced during rotations toward the lesion side in many animal species (Beraneck et al. 2008; Broussard et al. 1999; Crane and Demer 1998; Fetter and Zee 1988; Gilchrist et al. 1998; Halmagyi et al. 1990; Lasker et al. 2000; Newlands et al. 2005; Paige 1983; Sadeghi et al. 2006; Shinder et al. 2005; Vibert et al. 1993). Large deficits in the translational VOR (tVOR) were also reported several months postlesion in nonhuman primates (Angelaki et al. 2000).

Neural correlates of rVOR deficits and compensation have been found in the vestibular nuclei (VN), an area that receives direct vestibular afferent inputs and projects to both vestibuloocular and thalamocortical systems (Marlinski and McCrea 2008; Meng et al. 2007; Uchino and Kushiro 2011). In macaques, yaw rotation response gains of type I neurons (i.e., cells with firing rate increases during ipsilateral rotation) in the contralesional VN show nearly complete recovery 1 mo postlesion (Newlands et al. 2014; Newlands and Wei 2013; Sadeghi et al. 2011), much faster and more completely than in anesthetized or decerebrate animals (Newlands and Perachio 1990a; Smith and Curthoys 1988a, 1988b).

Despite this wealth of interest in vestibular compensation from unilateral peripheral damage, our knowledge remains incomplete for multiple reasons. First, studies after LB have nearly exclusively used yaw rotations only. VN neurons, however, respond to rotations in three dimensions (3D). Thus there is currently no data on 3D rotational gain recovery. Second, neural compensation in the otolith system has typically been tested in anesthetized animals (Chan 1997; Chan et al. 1999; Hoshino and Pompeiano 1977; Lacour et al. 1985; Xerri et al. 1983), and little is currently known about central recovery of otolith pathways in alert animals (Newlands et al. 2014). Third, few previous studies quantified neuronal variability in addition to mean responses, and those that did only measured thresholds for a short time after LB (Jamali et al. 2014; Liu et al. 2013a; Newlands et al. 2014). As a result, no information is currently available on whether spike count variability and neuronal thresholds are compensated fully following long-term recovery from unilateral vestibular damage.

Bilateral interactions are not only limited to the VN. Communication between the two labyrinths through the efferent system has been clearly demonstrated in birds, where mechanical stimulation of the horizontal canal on one ear affected afferent activity on the other side (Dickman and Correia 1993), and a role of the vestibular efferent system in long-term modulation of afferent responses has also been suggested. Thus it was perhaps surprising that macaque semicircular canal afferents in the intact (IN) nerve showed no change in response gain, phase, resting discharge and dynamic range, as well as other mean response properties following LB (Sadeghi et al. 2007b). The only significant change was an increase in the proportion of irregular afferents postlesion, although lesion-induced changes in the spike count variability of vestibular afferent responses were not examined (Sadeghi et al. 2007b).

Here we have measured spike count variability and direction detection thresholds of horizontal, anterior and posterior semicircular canal afferents, as well as VN neurons, in both IN and compensated unilaterally LB macaques. We found higher afferent neural rotation thresholds in LB animals, and this difference was traced down to increased trial-by-trial variability, whereas afferent mean response properties remained unchanged after contralateral vestibular lesion. In addition, we report persistent chronic deficits in 3D rotation detection thresholds for both ipsilesional and contralesional VN neurons. In contrast, translation thresholds of VN neurons recovered completely to normal values, suggesting that a single labyrinth is sufficient to generate highest sensitivity for the otolith, but not the semicircular canal, system.

METHODS

3D rotations were delivered using a six-degree-of-freedom motion platform (Moog 6DOF2000E) that was controlled using an Ethernet interface (60 Hz). Data were collected from five labyrinth-IN male rhesus monkeys (Macaca mulatta), which were chronically implanted with a plastic head-restraint ring and a guide tube platform, as described in detail in previous publications (Dickman and Angelaki 1999; Gu et al. 2006; Meng et al. 2005). One of these animals, plus a sixth macaque, were subjected to LB (animals C and W, right and left side, respectively; same animals as in Liu et al. 2013a). Additionally, a scleral search coil was implanted for monitoring eye movements (for details, see Gu et al. 2006; Meng et al. 2007). All surgical procedures were performed under sterile conditions and approved by the Institutional Animal Care and Use Committees at Washington University School of Medicine and Baylor College of Medicine, in accordance with the National Institutes of Health and Institutional guidelines.

Canal afferent responses were collected over a period of 5 to 8 mo after the vestibular lesion. We recorded extracellular neural activity from 238 canal afferents from IN monkeys (animal K: n = 91; animal H: n = 60; animal C: n = 87) and 161 canal afferents from the contralateral nerve after LB (animal C: n = 56; animal W: n = 105) using high-impedance, epoxy-coated tungsten microelectrodes (FHC, Bowdoinham, ME). The vestibular nerve was isolated beneath the auditory meatus as it entered the brain, often in close proximity to the cochlear nucleus (see Haque et al. 2004; Yu et al. 2012).

The VN were identified using the abducens nuclei with their characteristic firing patterns during eye movements as a landmark (Klier et al. 2006). VN penetrations were then concentrated 1–5 mm posterior and 0–3 mm lateral to the abducens nucleus. VN responses were collected over a period of 3 to 25 mo after the LB (animal C: n = 32 cells; animal W: n = 34 cells). These responses are compared with data from 43 cells from the IN macaques (animal K: n = 7; animal S: n = 19; animal T: n = 17).

Electrodes were inserted into 26-gauge transdural guide tubes and advanced by a remote-controlled microdrive (FHC). Raw neural activity was amplified, filtered and passed through a dual time-amplitude window discriminator (BAK Electronics). The spike train of each recorded cell was analyzed offline. Single units were identified based on waveform shape, latency, and amplitude. Only well-isolated neurons were included for further analyses.

Each VN neuron was first tested for eye movement-related activity. Responses during pursuit of a target moving in the four cardinal directions, up, down, left and right (Gaussian velocity profile, 9° amplitude and 2-s duration), were used to determine if a cell had significant eye movement-related tuning (P < 0.05, one-way ANOVA). Only VN cells without eye movement-related activity (“vestibular-only” cells) were used for further testing and analysis. All afferent and VN data were collected in darkness, and animals were always sitting upright relative to the motion platform and gravity, with the horizontal stereotaxic plane earth-horizontal. In this position, yaw rotation activates primarily (but not exclusively; e.g., Fig. 1, top) the horizontal semicircular canals (Haque et al. 2004). Upon isolation, the following stimuli were delivered.

Fig. 1.

Fig. 1.

Example anterior semicircular canal afferent response during the classification protocol. A cell's modulation during three cycles of sinusoidal rotation about the yaw, left-anterior-right-posterior (LARP) and right-anterior-left-posterior (RALP) axes are shown as instantaneous firing rate (IFR). Each dot corresponds to an IFR value. Stimuli (0.5 Hz) are shown below the IFR. Gray lines illustrate Kaiser filtered data (see methods) superimposed on top of the IFR. sp/s, Spikes per second; FB, forward-backward; LR, left-right.

1) Classification (CLA) protocol: The spontaneous activity with the animal upright and stationary was recorded first. Next, cells were tested with a “CLA protocol” to determine their preferred directions (PD); sinusoidal stimuli (0.5 Hz, 9–10 cycles, lasting ∼20 s) were delivered in three directions of rotation roughly aligned with the three semicircular canal axes [yaw, right-anterior/left-posterior (RALP) and left-anterior/right posterior (LARP), 22°/s peak amplitude] and in the three orthogonal directions of translation: lateral (LR; 0.1 G), nasooccipital (NO; 0.1 G) and dorsoventral (DV; 0.07 G). All three motion directions were successfully tested in 152 IN and 125 LB well-isolated canal afferent fibers and 42 IN and 64 LB (rotation) vs. 35 IN and 50 LB (translation) VN cells.

2) Variable-magnitude (VMAG) rotations: For canal-responsive cells (both canal afferents and VN neurons), the main experimental protocol consisted of 0.5-Hz sinusoidal rotations about the PD of each canal afferent. A block of VMAG consisted of ∼20 s of spontaneous activity (0°/s amplitude) and nine cycles of each of six different peak amplitudes (0.5°/s, 1°/s, 2°/s, 4°/s, 7.8°/s and 15.6°/s), randomly interleaved, lasting ∼3 min in total. By comparing peak and trough responses to these different amplitude stimuli, we could measure the smallest amplitude for which an ideal observer could discriminate the two rotation directions based on the firing rate of the particular neuron. All canal afferents included in the VMAG analysis (117 IN and 95 LB) were tested for at least two blocks (i.e., a minimum of 18 cycles for each rotation amplitude; lasting ∼6 min), the majority of neurons (60.4%) had four blocks (∼12 min) and 8.5% had five blocks (∼16 min). In a small subgroup of these recordings (n = 3), two semicircular canal afferents were simultaneously recorded for a minimum of two blocks.

For translation-responsive VN cells, translation detection thresholds were measured from responses to 0.005, 0.01, 0.02, 0.04, 0.08 and 0.16 G sinusoids (1 Hz) delivered along one or more of the same three directions as in the CLA protocol. The order in which the six different sets of stimuli (3D rotation and 3D translation) were delivered for VN cells was based on the outcome of the CLA protocol: the stimulus direction for which the cell was most responsive was tested first, the second-most responsive direction was recorded second, and so forth. If cell isolation was lost before the end of the entire protocol, only some of the six cardinal directions were included in the analysis. Note that we did not record from otolith afferents after LB. Otolith afferent 3D thresholds in IN animals using an identical experimental protocol have been presented in Yu et al. (2012).

3) 3D tuning protocol: Eleven pairs of canal afferents were also recorded simultaneously during a 3D tuning protocol, which consisted of transient rotations about 26 axes (that included pitch, yaw and roll), evenly spaced on a sphere (Chen et al. 2010; Takahashi et al. 2007). The stimulus trajectory followed a Gaussian velocity profile with a duration of 2 s, peak rotation amplitude was 9° and peak angular velocity was ∼20°/s. Five repetitions of each stimulus direction were tested (lasting ∼9 min). The 3D tuning data, along with the pair recordings during the VMAG protocol, were used to measure interneuronal correlations (see below). Note that this analysis was not performed for VN neurons because VN interneuronal correlations have been previously measured and reported (Liu et al. 2013b).

Data Analysis (Canal Afferents)

To classify canal afferents based on discharge regularity, the distribution of interspike intervals (ISIs) recorded during spontaneous activity was used to compute the coefficient of variation (CV), CV = σISIISI, where σISI and μISI are the SD and mean of the ISI distribution, respectively. Because CV varies with the mean ISI, we used a normalized CV (CV*) to classify vestibular afferents (Goldberg et al. 1984, 1990a, 1990b). In line with previous studies, neurons with CV* < 0.10 were classified as “regular” firing, whereas those with a CV* > 0.10 were classified as “irregular” canal afferents.

Sinusoidal responses during the 0.5-Hz oscillations about the yaw, LARP and RALP axes (CLA protocol) were first quantified using instantaneous firing rate (IFR), computed as the inverse of the ISI and assigned to the middle of the ISI (Fig. 1; each dot corresponds to one IFR value). We then stacked responses to repeated stimulus cycles into a single IFR cycle. Gain and phase were then calculated by fitting both neural response (clipped off at zero firing rate) and stimulus with a sine function (first and second harmonics and DC offset) using a nonlinear least squares algorithm (Levenberg-Marquardt method). Response amplitude was then measured to be one-half the peak-to-trough first harmonic modulation of the sinusoidal fit. Neural gain was computed as the ratio of response modulation amplitude over the stimulus velocity amplitude (in units of spikes·s−1 per °·s−1). These gain values along the three orthogonal directions were subsequently used to compute the 3D PD and the response gain along the PD based on a cosine-model for response tuning (Fernandez and Goldberg 1971; Goldberg and Fernandez 1971b).

To measure direction detection thresholds (i.e., the sensitivity of individual neurons to detect the direction of sinusoidal rotation), we compared distributions of firing rate (across multiple cycles of the VMAG protocol) for each trough and peak of response modulation along the PD of each cell; that is, yaw, RALP and LARP axis for horizontal, RALP and LARP canal afferents, respectively. The timing of response peak and trough (as response phase is generally non-zero) was computed by shifting the time of each zero-crossing of the stimulus according to the phase of the measured neural response using the phase estimated from the sinusoidal fit. Thus, for each cycle, we computed firing rate (spike count) within two temporal windows of 500 ms, each centered on trough and peak, that corresponded to opposite stimulus directions. Distributions of peaks and troughs for each stimulus amplitude were then subjected to ROC analysis to compute discriminability and construct neurometric functions (Gu et al. 2007, 2008; Yu et al. 2012). These neurometric functions were then fit with cumulative Gaussian functions, and neural threshold was defined as the standard deviation of the Gaussian (corresponding to 84% correct performance). These same data were also used to measure amplitude tuning slope and response (spike count) variability.

To illustrate that the actual threshold and variance measured depend on how neural firing rates are computed, we varied the width of the temporal window during which spike counts were measured at peak and trough. In addition, for comparison with a previous study (Sadeghi et al. 2007a), we also computed thresholds and variance after firing rate was converted into a continuous function using a low-pass filter with a Kaiser window and cutoff frequency greater than the stimulus frequency by 0.1 Hz (Sadeghi et al. 2007b; Fig. 1, gray lines). Using the Kaiser-filtered data, we also computed response variance using the smaller number of cycles of the CLA experimental protocol. Because only a single rotation amplitude was used for CLA, we computed firing rate distributions by binning each response cycle into bins of 1°/s, after shifting the time of each zero-crossing of the stimulus according to the phase of the measured neural response. These distributions were then used to compute response variability (defined as “CLA variance”), which was subsequently compared on a cell-by-cell basis with the VMAG variance for 97 IN and 82 LB canal afferents for which both datasets were available. Notice that, whereas response variance from the VMAG protocol was computed from data collected in a time frame of several minutes (6–18 min), CLA variance was measured using data collected within the time frame of no more than 20 s.

Data Analysis (VN Neurons)

For the VN data, detection thresholds were only measured for stimulus directions for which the cell showed significant modulation. This was determined using a permutation analysis, which was thus applied to firing rates during each of the six cardinal stimulus directions tested under the CLA protocol. For this, IFR responses were binned into 40 bins per cycle, and a Fourier ratio (FR) was defined as the fundamental frequency divided by the maximum of any of the first 20 harmonics. Then the 40 bins were shuffled randomly to remove any modulation, and a FR was calculated again. This randomization process was repeated 1,000 times, and, if the original FR was higher than the FRs of the 99% of the permuted data, the modulation was considered to be significant (P < 0.01). If a cell responded significantly to one or more directions of rotation or one or more directions of translation (but not to eye pursuit), the cell was used for further analysis. To calculate the cell's rotational and translational PD, responses during the CLA protocol were quantified by measuring the gain and phase for each of the three rotational directions and each of the three translational directions, as long as at least one of the three cardinal directions passed the permutation significance test. Gain and phase were computed by superimposing the IFR of each cycle (similarly as for vestibular afferents), and fitting the stacked IFR data with a sinusoidal function (Meng et al. 2005; Yu et al. 2012). We then used a spatiotemporal cosinelike model to compute 3D PD and gain (Angelaki 1991, 1992; Schor and Angelaki 1992). Direction detection thresholds for rotation were measured in VN neurons using the VMAG protocol in an identical analysis as the canal afferents. However, direction detection thresholds for translation were analyzed using 250-ms windows to ensure the window size was kept constant relative to the cycle duration (¼ of a cycle).

Interneuronal (“Noise”) Correlation Analysis

To analyze trial-by-trial correlations between firing rates of simultaneously recorded canal afferents, we searched for pairs of well-isolated single units from the same electrode during the VMAG and/or 3D tuning protocol using spike-sorting software. Such canal-canal cell pairs could be extracted from five recordings from IN animals and from nine recordings from LB animals. Interneuronal correlation was computed as the Pearson correlation coefficient (ranging between −1 and 1) of the normalized trial-by-trial responses from the spike counts during the 400 ms centered around peak and trough (sinusoidal protocols) or the middle 400 ms of the stimulus duration (3D tuning protocol) (Bair et al. 2001; Gu et al. 2011; Zohary et al. 1994). Because rotation direction and/or stimulus amplitude was varied across trials, spike counts from individual trials were z-scored to remove the stimulus effect and to allow pooling of data across different stimuli. Thus the measured correlation reflected only trial-to-trial variability. We then pooled data across different stimuli to compute one correlation value; the corresponding P value was used to assess the significance of correlation for each pair of neurons.

RESULTS

Semicircular Canal Afferents: Basic Response Characteristics and Comparison of Mean Responses

We recorded from 238 canal afferents bilaterally in IN animals and 161 canal afferents from the contralateral nerve in animals with one labyrinth surgically ablated (LB). Each fiber was first characterized in terms of its spontaneous activity and its modulation during 0.5-Hz oscillations along three axes: Yaw, LARP, RALP (Fig. 1). The basic response properties of IN semicircular canal afferents were similar to those described in previous studies (Bronte-Stewart and Lisberger 1994; Goldberg and Fernandez 1971a; 1971b; Haque et al. 2004; Ramachandran and Lisberger 2006; Sadeghi et al. 2006, 2007a, 2007b). The average resting firing rate was 96.0 ± 35.2 (SD) spikes/s, and the yaw response gain of horizontal canal afferents averaged 0.40 ± 0.19 spikes·s−1 per °·s−1 and 1.21 ± 0.54 spikes·s−1 per °·s−1 for regular and irregular afferents, respectively.

The PD gain and phase were computed for both IN and LB canal afferents, with response properties summarized in Figs. 2 and 3. Both response gain and phase showed a strong dependence on discharge regularity, quantified by the CV* (Goldberg et al. 1990a, 1990b). The more irregular the fiber's firing rate, the higher its PD gain (Fig. 2A) and the larger the phase lead relative to rotational velocity (Fig. 2B). Both of these relationships were significant (PD gain vs. CV*: r = 0.77, P < 0.001; phase vs. CV*: r = 0.84, P < 0.001, type II regressions). At 0.5 Hz, the mean canal afferent PD gain was (geometric means and SD) 0.42 ± 1.75 spikes·s−1 per °·s−1 and 1.35 ± 1.51 spikes·s−1 per °·s−1 for regular and irregular fibers, respectively (Fig. 2A, black symbols and bars). The average phase leads for regular and irregular afferents were 10.0 ± 6.1° and 27.2 ± 9.2°, respectively (Fig. 2B, black symbols and bars). There was no significant difference in either response gain or phase based on canal type [i.e., horizontal, anterior or posterior canal; analysis of covariance (ANCOVA), P = 0.92 for gain and P = 0.61 for phase].

Fig. 2.

Fig. 2.

Summary of mean response properties of semicircular canal afferents in intact (IN; black) and lesioned (LB; red) animals. A: response gain [along the three-dimensional (3D) preferred direction (PD)] is plotted as a function of normalized coefficient of variation (CV*) for n = 152 (IN) and n = 125 (LB) cells. The marginal histograms on the right show gain distributions (arrows illustrate geometric means) for regular (IN: 122, LB: 85) and irregular (IN: 30, LB: 40) afferents. B: response phase as a function of CV* for n = 238 (IN) and n = 161 (LB) canal afferents. Marginal histograms on the right show phase distributions (arrows illustrate geometric means) for regular (IN: 180, LB: 112) and irregular (IN: 58, LB: 49) afferents. Different symbols in the scatterplots illustrate different canal types: circles, horizontal (H) canals; squares, anterior (A) canals; and diamonds, posterior (P) canals. The difference in the number of cells between A and B is due to loss of cell isolation before the classification (CLA) protocol was completed (thus, a PD gain could not be measured). C: distributions of CV* (arrows illustrate geometric means). D: cumulative distributions of CV* for IN (black) and LB (red) afferents (IN: 238 and LB: 161).

Fig. 3.

Fig. 3.

Summary of PD. Each unit vector is shown in stereotaxic coordinates (front, top, and left side views; see drawing) according to the right-hand rule. Red circles, green triangles, and blue diamonds represent H, A, and P afferent vectors, respectively. LAC: left anterior canal; LHC: left horizontal canal; LPC: left posterior canal; RAC: right anterior canal; RHC: right horizontal canal; RPC: right posterior canal. Black lines: IN afferents (n = 152); gray lines: LB afferents (n = 125). Dotted lines illustrated the 3 axes used to test canal afferents during the CLA protocol.

There was also no significant difference in PD gain for IN vs. LB afferents (regular: P = 0.6; irregular: P = 0.08; Fig. 2A, marginal histograms) or phase (P = 0.6 for both regular and irregular afferents; Fig. 2B, marginal histograms). In addition, we found no difference between IN and LB datasets in how PD gain and phase varied with frequency or canal type (ANCOVA, P > 0.3). Note that we did not encounter many cells with high CV* and low-response gains, which have been shown to correspond to canal afferents forming calyx endings in the central striola region in rodents (C-irregular; Baird et al. 1988). The absence of such units in our sample is also shown by the absence of a clear second peak in the CV* distribution (Fig. 2C) (Baird et al. 1988; Fernandez and Goldberg 1971). In Fig. 2, A and B, we have marked with filled symbols the few canal afferents with CV* > 0.2 and gain < 1, equivalent to the criterion used by Sadeghi et al. (2007a) to distinguish putative calyx-type from dimorph-type irregular canal afferents.

In addition to no change in mean gain and phase, we also found no significant difference in spontaneous firing rate for IN [94.6 ± 30.8 (SD) spikes/s] vs. LB (98.9 ± 31.1 spikes/s) afferents (P = 0.23, Wilcoxon rank test). However, the percentage of irregular afferents increased after LB (30% vs. 24%). The geometric mean of the corresponding CV* distributions [± geometric SD: 0.05 ± 2.34 (IN) vs. 0.06 ± 2.78 (LB)] were different after LB (P = 0.04, Wilcoxon rank test; Fig. 2C, compare red and black distributions). Such difference in CV* can be further illustrated by comparing the cumulative distributions (Fig. 2D, IN: black, LB: red). A similar finding was reported by Sadeghi et al. (2007a).

PD are shown as unit vector projections onto the three cardinal planes in Fig. 3 (IN: black lines; LB: gray lines). Note that PDs do not exactly coincide with the stimulus axes because animals were positioned in stereotaxic coordinates, i.e., the yaw rotation plane was the stereotaxic horizontal rather than the plane of the horizontal semicircular canals (Fig. 3, dashed lines; see also Haque et al. 2004). Nevertheless, as the difference is small (and identical for IN and LB afferents), the remaining analyses have used the maximum response direction (yaw for horizontal canal afferents, RALP for right anterior and left posterior afferents and LARP for left anterior and right posterior afferents) to quantify neural threshold and variance.

Semicircular Canal Afferents: Direction Detection Thresholds

The computation of the direction detection threshold for an example left anterior canal afferent (CV* = 0.3) is shown in Fig. 4. We also illustrate the smooth, continuous function generated using the Kaiser filter (Fig. 4A, gray curve superimposed on IFR; see also Fig. 1). From spike count distributions created by quantifying peak and trough firing rates (500-ms temporal window), the sensitivity of each cell for discriminating oppositely directed rotations was quantified using signal detection theory (see methods). The firing rates were first converted into a two-alternative forced choice format using 500-ms bins centered around the peak and trough of each response cycle (Fig. 4A). We then grouped together all firing rates within each peak (filled bars) and trough (open bars), for all stimulus magnitudes (0.5–15.6°/s; Fig. 4B). For the example cell shown, distributions are far apart for 15.6°/s and 7.8°/s, have a very small overlap at 2°/s and completely overlap at 0.5°/s (Fig. 4B). As a result, an ideal observer would be able to distinguish whether a given firing rate came from one or the other distribution with a high degree of certainty for rotations larger than 2°/s, but not for 0.5°/s. Thus the neuron's direction detection threshold should be between 0.5°/s and 2°/s. Indeed, as illustrated by the afferent's neurometric function (Fig. 4C), this neuron had a direction detection threshold σ = 1.37°/s.

Fig. 4.

Fig. 4.

Computation of direction detection threshold for an anterior canal afferent. A: IFR of the same afferent as in Fig. 1 with superimposed gray line illustrating the Kaiser filter response (see methods), as well as corresponding raster plots for different trials during rotation about the cell's maximum response direction. Bars mark 500-ms intervals used to compute mean firing rate for peak (filled bar) and trough (open bar) response, analyzed cycle by cycle. B: firing rate distributions for 4 pairs of stimulus magnitudes (15.6°/s, 7.8°/s, 2.0°/s, and 0.5°/s), with filled bars corresponding to peak, and open bars corresponding to trough spike counts. C: example neurometric function, with circles showing proportion direction decisions of an ideal observer as a function of peak velocity magnitude, computed from distributions like those shown in B. Solid lines show cumulative Gaussian fit to the neurometric function.

Direction detection thresholds for the entire macaque canal afferent population are summarized in Fig. 5. When all cells were considered together, there was only a weak dependence of neural threshold on CV* [ANCOVA, P = 0.05; geometric mean ± SD: 2.83 ± 1.42°/s (regular, IN), 2.00 ± 1.48°/s (irregular, IN)]. However, when putative calyx afferents were excluded, this dependence became stronger (P < 0.001; irregular: 1.85 ± 1.37°/s). A similar conclusion was also reached for otolith afferents (Yu et al. 2012). We found no difference in neural threshold for horizontal vs. vertical canal afferents (ANCOVA main effect, P = 0.74; Fig. 5, different symbols; see also Table 1), so data from all three canal afferent types have been pooled together for subsequent analyses.

Fig. 5.

Fig. 5.

Summary of direction detection thresholds. Neural threshold is plotted as a function of CV* on a cell-by-cell basis, shape-coded according to canal type (H: circle, A: square, P: diamond) and color-coded according to IN (black) vs. LB (red) type. Marginal histograms on the right show threshold distributions, separately for IN (black, n = 117) and LB (red, n = 95) data (arrows illustrate geometric means). Filled symbols: putative C-irregular neurons.

Table 1.

Rotation detection thresholds for semicircular canal afferents

Intact
LB
All Regular Irregular All Regular Irregular
Horizontal semicircular canals 2.81 ± 1.49 (36) 3.34 ± 1.37 (21) 2.21 ± 1.45 (15) 3.47 ± 1.73 (36) 3.52 ± 1.56 (23) 3.38 ± 2.00 (13)
Anterior semicircular canals 2.40 ± 1.50 (36) 2.59 ± 1.45 (30) 1.66 ± 1.48 (6) 3.17 ± 1.87 (34) 3.52 ± 1.73 (27) 2.13 ± 2.09 (7)
Posterior semicircular canals 2.54 ± 1.44 (45) 2.77 ± 1.37 (34) 1.92 ± 1.45 (11) 3.06 ± 1.70 (25) 3.65 ± 1.39 (16) 2.24 ± 1.94 (9)
All 2.57 ± 1.48 (117) 2.83 ± 1.42 (85) 2.00 ± 1.48 (32) 3.25 ± 1.78 (95) 3.55 ± 1.59 (66) 2.66 ± 2.07 (29)

Values are geometric means ± SD separated by canal type (rows) and intact vs. labyrinthectomy (LB, columns). Numbers in parentheses, number of canal afferents tested.

Even though there was no difference in mean response gain and phase, we found that LB canal afferent thresholds were significantly higher than IN afferent thresholds (geometric mean ± SD: 3.25 ± 1.78°/s vs. 2.57 ± 1.48°/s, i.e., a 27% increase; Wilcoxon test, P < 0.001; Table 1). This difference reached significance even when comparing the thresholds for the one animal tested both before and after LB [IN: 2.74 ± 1.46°/s (n = 51) vs. LB: 3.48 ± 1.82°/s (n = 42); Wilcoxon test, P = 0.02]. Note that the few putative calyx-type afferents had high thresholds and likely large differences between IN and LB animals (4.06 ± 1.38°/s vs. 7.23 ± 1.24°/s, compared with 2.83 ± 1.42°/s vs. 3.55 ± 1.59°/s for regular afferents and 1.85 ± 1.37°/s vs. 1.94 ± 1.67°/s for irregular-dimorph afferents), although the small number of putative calyx units encountered prevented further quantification.

Semicircular Canal Afferents: Increased Variability in LB vs. IN Canal Afferents

Neural thresholds depended on two properties that determine how much the two spike count distributions overlap (Fig. 4B): the difference in their means and their width (variance) (ANCOVA main effects, P < 0.001). The difference in the distribution means is directly related to the slope (steepness) of the magnitude tuning curve, a parameter that is similar to response gain [because all magnitude tuning curves were linear, with a mean correlation coefficient of r2 = 0.97 ± 0.003 (SD) (IN) and r2 = 0.95 ± 0.008 (LB)]. The dependence of neural threshold on the magnitude slope and variance (computed for each cell as the median for all tested velocities) was identical in LB and IN datasets (ANCOVA, P > 0.14; Fig. 6, A and B). Furthermore, there was no difference in magnitude slope between IN and LB afferents (geometric mean ± SD: 0.58 ± 2.08 spikes·s−1 per °·s−1 vs. 0.57 ± 2.37 spikes·s−1 per °·s−1, respectively; Wilcoxon rank test, P = 0.86; Fig. 6A, red vs. black histograms), in line with the fact that response gains were also unchanged (Fig. 2). Instead, the threshold difference between IN and LB afferents was due to differences in the spike count variance, which was significantly larger in LB (geometric mean ± SD: 6.78 ± 3.50 spikes2/s2) than in IN (4.46 ± 2.99 spikes2/s2) animals (Wilcoxon rank test, P = 0.01; Fig. 6B, red vs. black histograms).

Fig. 6.

Fig. 6.

Parameters influencing neural threshold. A and B: dependence of threshold on magnitude tuning curve slope (A, in units of sp·s−1 per °·s−1) and median variance [B, in units of (sp/s)2]. Median variance is computed from multiple response amplitude distributions, like those in Fig. 4B. Histograms on the top show the corresponding distributions of magnitude tuning curve slope and median variance (arrows illustrate geometric means). Magnitude tuning curve slope (C) and median variance (D) plotted as a function of CV*. Black symbols: IN (n = 117). Red symbols: LB (n = 95). Solid lines show type II linear regression (plotted only whenever correlation was significant). Filled symbols: putative C-irregular neurons.

When plotted as a function of CV*, both magnitude slope and variance scaled with CV* (Fig. 6, C and D; type II regression, P < 0.001 for both IN and LB data). That is, irregular afferents had both higher gain and higher variance than regular afferents, such that irregular afferents had similar or even smaller thresholds compared with regular canal afferents. A similar property was also a notable characteristic of otolith afferents (see Fig. 5 of Yu et al. 2012): magnitude slope and variance both increased similarly, if not more, with increasing CV*, such that irregular afferents were generally as or more sensitive than regular afferents. Note though that putative calyx canal fibers (Fig. 6, filled symbols) might not share this property, as their gains do not follow the monotonic dependence on CV* (see discussion). Due to the small number, putative calyx canal afferents have not been distinguished in subsequent plots.

The difference in spike count variance between LB and IN afferents has been further illustrated in Fig. 7, where the data are plotted as a function of magnitude slope. For IN afferents, there was a strong correlation between the two parameters (r = 0.78, P < 0.001, type II regression). The higher the magnitude slope (i.e., gain), the larger the cell's variance, thus resulting in thresholds that do not show a strong dependence on CV* (Fig. 5). This relationship was weaker, but still significant in LB afferents (r = 0.43, P < <0.001). Notably, the relationship between spike count variance and magnitude slope was significantly altered after lesion of the contralateral labyrinth, as illustrated in Fig. 7 [red vs. black type II regression: slope 95% confidence interval (CI): (19.7, 32.6) vs. (36.8, 138.1); intercept 95% CI: (−15.3, −7.3) vs. (−97.9, −17.4)]. Thus, for the same magnitude slope (i.e., response gain), spike count variance was higher in LB than in IN animals. The difference in spike count variance between LB and IN afferents was also significant when afferents were separated into regular (P = 0.014, Wilcoxon rank test) and irregular (P = 0.015) groups (Fig. 7, left marginal histograms). In contrast, there was no difference in magnitude slope for LB vs. IN data (P > 0.38; Fig. 7, bottom marginal histograms).

Fig. 7.

Fig. 7.

Dependence of median variance on magnitude tuning curve slope. Median variance is computed for each direction from multiple response amplitude distributions, like those in Fig. 4B. IN: n = 117 (black); LB: n = 95 (red). Solid lines: type II linear regression. Marginal histograms show slope distributions (bottom) and median variance distributions (left) for regular (IN: 85, LB: 66) and irregular (IN: 32, LB: 29) afferents separately (arrows illustrate geometric means).

As our laboratory has previously reported (Yu et al. 2012), the exact threshold (Fig. 8A) and spike count variance (Fig. 8B) measured depends on the filtering method (e.g., Kaiser filter vs. spike count) and window size used to compute the opposite direction response distributions (Fig. 4B, filled vs. open bars). In contrast, magnitude slope was largely independent of both filtering method and window size (Fig. 8C). Thus one cannot compare neural thresholds from different studies using different firing rate measures. Also, it is inappropriate to directly compare neural thresholds in macaques with behavioral thresholds in humans (see discussion), so we have not attempted such comparisons here. Importantly, differences between IN and LB threshold (Fig. 8A) and variance (Fig. 8B) persisted, regardless of the filtering method used to quantify the firing rate distributions. Thus, although the exact magnitude of measured neural threshold should be interpreted with caution, the measured differences between IN and LB afferents hold independently of the analysis method used.

Fig. 8.

Fig. 8.

Effects of filtering method used for firing rate computation. Average detection threshold (A), median variance (B), and magnitude tuning curve slope (C) plotted as a function of window size for two different filtering methods: spike count (solid lines) and Kaiser filter (dotted lines). Black: IN; red: LB. Error bars represent standard error.

The VMAG used here to compute neural threshold took several minutes (6–18 min) to complete (see methods). Thus it is possible that the observed increase in cycle-by-cycle variability might not be present when the variance is computed for shorter time intervals. To investigate the time scale of increased variability, we compared response variance computed from the VMAG (6–18 min) and CLA (20 s) protocols (see methods). For the latter, following the analysis of Sadeghi et al. (2007b), head velocity was binned into 1°/s steps, and the analysis illustrated in Fig. 4 was performed on distributions centered on those bins.

The median firing rate variability computed from the short CLA protocol also increased in LB compared with IN afferents (geometric mean ± SD, averaged over all velocity bins: 3.50 ± 4.90 spikes2/s2 vs. 2.51 ± 3.72 spikes2/s2; P < 0.001, Wilcoxon rank test). However, the VMAG variance was larger than the CLA variance, as illustrated by most data points falling above the unity-slope dotted line in the scatter plot of Fig. 9, A and B (P < <0.001, Wilcoxon sign rank test). In addition, the difference between these two variability measures, computed as a “VMAG/CLA variance ratio,” increased after LB (1.37 ± 1.17 vs. 2.73 ± 1.27, geometric mean and SD; P < 0.001, ANOVA; Fig. 9C). Thus, although LB vs. IN variability differences were significant for both CLA and VMAG datasets, the difference was larger when variance was computed over the time scale of minutes (VMAG protocol) than seconds (CLA protocol).

Fig. 9.

Fig. 9.

Relationship between firing rate variability computed from the CLA and variable magnitude (VMAG) protocols. A and B: scatterplots for IN (A) and LB (B) afferents, where each data point represents variability computed for one velocity bin for each neuron (13 data points per cell; note results are independent of the velocity bin; P = 0.16, ANOVA). Data are color-coded for IN (black, n = 97) vs. LB (red, n = 82) animals. Solid line: type II linear regression. C: distributions of VMAG/CLA variance ratios for IN (top) and LB (bottom) data (arrows illustrate geometric means). To exclude the possibility that the observed difference could arise due to the different trial numbers, we also measured variance separately for each of 13 stimulus amplitudes (0°/s, ±0.5°/s, ±1°/s, ±2°/s, ±4°/s, ±7.8°/s and ±15.6°/s) by randomly drawing 10 sample cycles from the VMAG protocol, and the results were similar.

Semicircular Canal Afferents: Interneuronal Correlations

The observed increase in firing rate variability in LB vs. IN semicircular canal afferents could reflect changes in interneuronal trial-by-trial (i.e., cycle-by-cycle) correlations. Multielectrode recordings from the vestibular nerve were not possible, but we could spike-sort offline a limited number of semicircular canal afferent pairs recorded from the same electrode. Thus, to investigate whether the observed increase in variance for LB afferents reflects correlated variability, we measured interneuronal correlations (i.e., the trial-by-trial correlations between the spike counts of two simultaneously recorded afferents) by spike-sorting records for which two single fibers could be distinctly discriminated offline.

Interneuronal correlations were computed as the Pearson correlation coefficient of the normalized trial-by-trial spike counts between two simultaneously recorded cells (see methods). Interneuronal correlations were measured for five (IN) and five (LB) pairs of afferents from the same canal and four (LB) afferents from different canals. Despite the limited dataset, correlations were larger in LB than IN afferents (0.17 ± 0.08 vs. −0.05 ± 0.10, mean ± SD; P = 0.002, Wilcoxon rank test), and this result did not appear to depend on CV* (Fig. 10). Thus the increased spike count variance observed in the intact vestibular nerve after LB could reflect correlated variability, although a larger number of afferent pairs would be needed to verify this finding.

Fig. 10.

Fig. 10.

Interneuronal correlations from IN (black, n = 5) and LB (red, n = 9) afferents are plotted vs. the CV* of the recorded pair of afferents (connected by line). Circles: same canal pairs; squares: different canal pairs. Filled symbols denote significant correlations (P < 0.05), whereas open symbols illustrate nonsignificant correlations. Each data set consisted of a minimum of 87 (and up to 540) trials.

VN Neurons: Basic Response Characteristics and Comparison of Mean Responses

We also analyzed responses from 109 VN neurons, of which 43 cells were recorded from three IN animals and 66 cells from the same 2 LB animals (35 LB-ipsi, i.e., cells recorded from the side ipsilateral to the lesion and 31 LB-contra, i.e., cells from the contralesional side). The majority of cells (82/109, 75%) showed significant modulation for at least one of the three cardinal directions during both rotation and translation (permutation test, P < 0.01, see methods). The remaining cells were significantly modulated during rotation only (24/109, 22%) or translation only (3/109, 3%). This ratio was consistent across the three groups of cells (IN: 34/43, 79%; 8/43, 19%; 1/43, 2%; LB-ipsi: 26/35, 74%; 8/35, 23%; 1/35, 3%; LB-contra: 22/31, 71%; 8/31, 26%; 1/31, 3%). None of the recorded cells showed any slow eye movement modulation, as determined during pursuit eye movements (see methods).

VN neurons were tested during three directions of rotation (yaw, LARP, RALP) and three directions of translation: LR, NO and DV, as illustrated with an example cell in Fig. 11, A and B. From these responses, 3D PD, gain and phase were computed using a spatio-temporal convergence model (see methods and Angelaki 1992; Schor and Angelaki 1992; data not shown). The distributions of spontaneous activity, as well as 3D rotation and translation response gains are summarized in Fig. 12. There was no difference in spontaneous activity for IN, LB-ipsi and LB-contra responses [35.7 ± 21.3 (SD), 35.9 ± 20.7, 41.6 ± 23.4 spikes/s; 3-way ANOVA, P = 0.45]. For cells that modulated significantly during rotation, response gains along the PD in 3D were lower after LB (3-way ANOVA, P = 0.002). Nonparametric comparisons revealed significant differences between LB-ipsi and both IN and LB-contra data (Wilcoxon's test, P < 0.001 and P = 0.001, respectively, see ** and * in Fig. 12B).

Fig. 11.

Fig. 11.

Example vestibular nucleus neuron (IN animal). A and B: CLA responses (IFR) using 0.5-Hz stimulation around three axes of rotation (yaw, LARP and RALP) and three directions of translation [lateral (LR), nasooccipital (NO), dorsoventral (DV)]. The solid gray lines show sinusoidal fits. This cell showed significant responses for yaw and RALP rotation (P < 0.001), but not for translation. C: computation of rotation thresholds using different amplitude sinusoids (0.5 Hz). Each dot corresponds to an IFR value. Gray lines illustrate Kaiser filtered data. Bars mark 500-ms intervals used to compute mean firing rate for peak and trough responses. D: corresponding spike count distributions centered on peak (black bars) and trough (white bars). E: velocity tuning curve. F: neurometric function, from which direction detection threshold was computed as the standard deviation of a cumulative Gaussian function fit.

Fig. 12.

Fig. 12.

Basic response properties of rotation- and translation-modulating cells. A: distribution of spontaneous firing rates, with arrows illustrating mean values (n = 109 cells). B and C: distributions of rotation and translation response gains along the PD in 3D. Arrows illustrate geometric means for data from IN (black, n = 42 for rotation and n = 35 for translation), LB-contralateral (hatched, n = 30 for rotation and n = 23 for translation) and LB-ipsilateral (hatched, n = 34 for rotation and n = 27 for translation) animals. Asterisks illustrate statistically significant mean differences (*P < 0.01; **P < 0.001, Wilcoxon's test).

In contrast, there was no difference in translation response gain computed along the 3D PD (IN: 213.7 ± 1.8, LB-contra: 190.9 ± 1.9, LB-ipsi: 182.1 ± 2.1 spikes·s−1G−1; 3-way ANOVA, P = 0.82). Since VN responses before and after unilateral vestibular lesion have never before been tested in 3D, no direct comparisons with previous studies are possible. However, considering only yaw rotation responses, we found significant gain differences between IN and LB animals for type I cells but not for type II cells (3-way ANOVA, type I: P = 0.04, n = 44; type II: P = 0.10, n = 45).

VN Neurons: Direction Detection Thresholds for Rotation and Translation

VN neurons were further characterized with stimuli consisting of sinusoidal rotations (0.5 Hz) or translations (1 Hz) along the cardinal axes at six different amplitudes (VMAG experimental protocol). Each cell was first tested in the cardinal direction with the largest response, followed by the other directions (see methods). Distributions of spike counts (Fig. 11, CE) then underwent ROC analysis to generate neurometric functions and measure neuronal threshold (Fig. 11F).

Rotation thresholds.

Yaw rotation thresholds were measured in 35/109 (32%) cells, LARP rotation thresholds in 44/109 (40%) cells and RALP rotation thresholds in 40/109 (37%) cells. For neurons recorded in IN animals, rotation thresholds averaged 6.9 ± 2.9°/s (geometric mean ± SD). Thresholds in chronically LB animals were significantly higher, 10.4 ± 2.2°/s (LB-contra) and 16.2 ± 2.3°/s (LB-ipsi) (Fig. 13A; ANOVA, P = 0.011). Post hoc analysis showed this significant difference was driven by the comparison between LB-ipsi and IN groups (P = 0.02, Tukey's honestly significant difference test for unequal sizes). Nonparametric comparisons revealed significant differences between IN and both LB-ipsi and LB-contra data threshold (Wilcoxon's test, P < 0.001 and P = 0.03, respectively, see * and ** in Fig. 13A, marginal histograms). When only yaw responses were considered, significant differences were found only for LB-ipsi vs. IN thresholds (type I: P = 0.008; type II: P = 0.04). In contrast, LB-contra yaw thresholds were not significantly different from those in IN animals (type I: P = 0.28; type II: P = 0.25).

Fig. 13.

Fig. 13.

Direction detection thresholds for rotation. A: neuronal thresholds as a function of the absolute difference between the tested direction and the preferred 3D direction of the cell, Δ(3D-PD), with data from each cell shown for 1–3 motion directions. Vertical dashed line marks Δ(3D-PD) = 90° (i.e., tested direction perpendicular to the cell's PD). Marginal histograms on the right show threshold distributions, separately for IN (black, n = 43), LB-contralateral (green, n = 41) and LB-ipsilateral (red, n = 35) animals (arrows illustrate geometric means). B: data in A folded around 90°, such that type II linear regression (solid) and inverse cosine function (dashed) lines are fit to the data, separately for the three groups of animals (values in Table 2). C: dependence of neuronal threshold on rotation magnitude tuning curve slope, with data from each cell shown for 1–3 motion directions. Different symbols are used for different animals (triangles: animal S; circles: animal T; squares: animal K; stars: animal C; diamonds: animal W).

As illustrated in the scatter plot of Fig. 13A, neuronal thresholds depended on the direction of the stimulus relative to the PD of maximal response, Δ(3D-PD), for each cell. The lowest thresholds were observed for rotation axes parallel to the PD [Δ(3D-PD) = 0° or 180°], and the largest thresholds were observed perpendicular (90°) to the PD (dashed vertical line in Fig. 13A). Thus, to estimate thresholds for stimulation along the cell's PD, data were folded along the dashed vertical line, such that the relationship was monotonic. Thresholds were significantly different (Fig. 13B; 3-way ANCOVA, P < 0.001), with larger deficits for LB-ipsi (P < 0.001, Post hoc Tukey's honestly significant difference test for unequal sizes) than LB-contra neurons (P = 0.04, comparison with IN group) compared with IN animals. Furthermore, LB-ipsi and LB-contra thresholds were significantly different (P = 0.04). This relationship was also quantified using a type II regression (Fig. 13B), whose intercept provided an estimate of the PD population threshold (Table 2). PD thresholds (±95 CI) were 2.5 (1.5–4.3), 3.4 (2.0–5.5) and 6.8 (4.9–10.3)°/s for IN, LB-contra and LB-ipsi groups, respectively. Note that inverse cosine fits gave slightly larger PD thresholds for all three groups, although fits were not better than those of a type II regression (Table 2).

Table 2.

VN detection thresholds for rotation

VN-Intact VN-LB-Contralateral VN-LB-Ipsilateral
Linear fit 2.5 (1.5–4.3) 3.4 (2.0–5.5) 6.8 (4.9–10.3)
r2 = 0.30 r2 = 0.43 r2 = 0.39
P < 0.001 P < 0.001 P < 0.001
Inverse cosine fit 4.2 (3.2–6.1) 5.8 (4.6–7.4) 9.6 (7.0–14.4)
r2 = 0.26 r2 = 0.49 r2 = 0.34
P = 0.001 P < 0.001 P < 0.001
Mean ± SD 6.9 ± 2.9 10.4 ± 2.2 16.2 ± 2.3

VN, vestibular nuclei. Top and middle rows: Neuronal thresholds in °/s with 95% confidence intervals (in parentheses), interpolated along the preferred direction from linear fits or cosine fits. Bottom row: geometric means and SD of thresholds along all tested directions (from distributions in Fig. 13A).

Neuronal thresholds depended on both the slope of the response amplitude tuning curve (e.g., Fig. 11E) and response variance (two-covariate ANCOVA, P < 0.001). Furthermore, the relationship between threshold and variance was consistent in IN and LB animals as well as the relationship between threshold and response amplitude tuning curve (interaction term, P = 0.60 and P = 0.65, respectively).

Translation thresholds.

LR stimulation was tested in 43/109 (39%) cells, NO stimulation was tested in 48/109 (44%) cells and DV stimulation was tested in 32/109 (29%) cells. In IN animals, translation thresholds averaged [geometric mean ± SD: 42.1 ± 3.4 cm/s2 (0.043 G)], which is not significantly different than otolith afferent thresholds (57.3 ± 2.8, P = 0.09, n = 97; Yu et al. 2012). Chronically LB animals had similar translation thresholds (LB-contra: 39.4 ± 3.1 cm/s2; LB-ipsi: 56.8 ± 2.9 cm/s2) as controls (3-way ANOVA, P = 0.85; Fig. 14, marginal histograms). Similar conclusions (i.e., no significant difference between IN and lesioned thresholds) were reached when accounting for the threshold dependence on the difference of the stimulus direction with the 3D-PD (ANCOVA, P = 0.27). Based on a type II regression, PD thresholds averaged (±95 CI) 13.8 (3.7–39.1), 18.7 (7.3–36.6) and 14.8 (6.4–34.2) cm/s2 for IN, LB-contra and LB-ipsi, respectively. Slightly higher PD thresholds were estimated using inverse cosine fits (Table 3). For comparison, otolith afferent thresholds along the 3D PD were 11.6 (6.8–18.1) and 22.7 (18.4–28.3) cm/s2, for linear and inverse cosine fits, respectively (considering regular and irregular afferents together; Yu et al. 2012). Regardless of the interpolation method, 95% CIs overlap, demonstrating that there is no difference in translation detection thresholds for either otolith afferents and VN neurons or IN and lesioned responses.

Fig. 14.

Fig. 14.

Direction detection thresholds for translation. Neuronal thresholds as a function of the Δ(3D-PD), folded around 90°, such that type II linear regression (solid color) and inverse cosine function (dashed) lines are fit to the data, are shown separately for the three groups of animals (data from each cell shown for 1–3 motion directions). Gray solid/dashed lines show corresponding fits for otolith afferents (data from Yu et al. 2012). Marginal histograms on the right show threshold distributions, separately for IN (black, n = 43), LB-contralateral (green, n = 31) and LB-ipsilateral (red, n = 49) animals (triangles: animal S; circles: animal T; squares: animal K; stars: animal C; diamonds: animal W).

Table 3.

VN detection thresholds for translation and comparison with otolith afferents

Otolith Afferents VN-Intact VN-LB-Contralateral VN-LB-Ipsilateral
Linear fit 11.6 (6.8–18.1) 13.8 (3.7–39.1) 18.7 (7.3–36.6) 14.8 (6.4–34.2)
r2 = 0.41 r2 = 0.13 r2 = 0.06 r2 = 0.23
P < 0.001 P = 0.019 P = 0.19 P < 0.001
Inverse cosine fit 22.7 (18.4–28.3) 20.0 (11.6–38.4) 27.4 (16.6–47.2) 25.9 (17.7–41.2)
r2 = 0.48 r2 = 0.18 r2 = 0.05 r2 = 0.28
P < 0.001 P = 0.005 P = 0.10 P < 0.001
Mean ± SD 57.3 ± 2.8 42.1 ± 3.4 39.4 ± 3.1 56.8 ± 2.9

Top and middle rows: Neuronal thresholds in cm/s2 with 95% confidence intervals (in parentheses), interpolated along the preferred direction from linear fits or cosine fits. Bottom row: Geometric means and SD of thresholds along all tested directions (from distributions in Fig. 14).

Finally, translation thresholds depended on both the amplitude tuning curve slope and variance (two-covariate ANCOVA, P = 0.002 and P = 0.03, respectively), and these relationships were not different in IN vs. LB animals (interaction terms: P = 0.62 and 0.54, respectively). Thus both response gain and variance, as well as translation detection thresholds appeared fully recovered after destruction of one labyrinth. Consequently, the remaining labyrinth is sufficient after compensation to maintain translation detection thresholds at the low values of IN animals.

DISCUSSION

We have characterized neural variability and direction detection thresholds of semicircular canal afferents and central VN neurons in IN macaques and animals whose contralateral labyrinth was surgically ablated. Unlike past studies that quantified mean responses only (Sadeghi et al. 2007b), here we show that spike count variance, and thus detection thresholds, of canal afferents were both significantly increased after contralateral labyrinth lesion. In agreement with Sadeghi et al. (2007a), we found no difference in resting discharge, as well as mean response gain and phase (Fig. 2). Sadeghi et al. (2007a) did report a decrease in the proportion of regular afferents (45 vs. 38%) after LB, similar to findings in the present study (76% vs. 70%). We also found a small (27%) but significant increase in the detection threshold of vestibular afferents after lesion of the contralateral labyrinth. Such increase in neural threshold was not due to differences in mean response gain, which remained unchanged, but rather due to a significant increase in spike count variability. Analysis of trial-by-trial spike count correlations of a small number of simultaneously recorded pairs of afferents revealed increased interneuronal correlations after LB. Although the number of simultaneously recorded canal afferents was low to support firm conclusions, increases in the measured correlations support the hypothesis that LB might increase correlated variability. The observed increase in variability could also be related to the higher mean CV* after LB (Fig. 2, C and D).

Importantly and for the first time after vestibular damage, VN neurons were tested using 3D rotations and translations. We note that VN thresholds interpolated along the cell's PD were similar to vestibular afferent thresholds (Tables 13). Because many VN cells responded during both rotation and translation, and because we tested cells during both yaw and vertical plane oscillations, VN rotation thresholds do not necessarily reflect activation of semicircular canals exclusively. For this reason, we made no attempt to compare VN and semicircular canal afferent thresholds for rotation, although we did compare VN and otolith afferent thresholds for translation (Fig. 14, data from Yu et al. 2012). In general, our analysis has shown much lower thresholds for both afferents and VN neurons than those reported previously by Cullen and colleagues (Jamali et al. 2013, 2014; Massot et al. 2011; Sadeghi et al. 2007a). One major difference is the fact that Massot et al. (2011), Sadeghi et al. (2007a) and Jamali et al. (2014) used stimuli limited to yaw rotations. Similarly, Jamali et al. (2013) used linear accelerations limited to the horizontal plane. In contrast, we have tested rotation and translation responses in 3D, which could at least partly explain the lower thresholds. However, we note that, as illustrated in Fig. 8, neural threshold depends on filtering and analyses used. Furthermore, comparisons between regular and irregular afferent thresholds can be skewed because of sampling biases and arbitrary cut-offs (e.g., CV* = 0.1). Thus comparisons between different laboratories using different motion platforms and different analyses might not be useful and justifiable.

Regardless, comparison among areas using identical stimuli and analyses, as done here, is extremely valuable. Remarkably, although convergence between the two labyrinths improved the modulation gain of central vestibular neurons in the VN (at least compared with regular afferents; compare Figs. 12B and 2A), it does not improve detection thresholds, and this is true for both rotation and translation stimuli (Tables 13). This likely occurs because of increased response variability in VN vs. vestibular afferents. Thus it is unclear whether there are any benefits of commissural convergence in sensory detection and discrimination of vestibular motion signals, and this hypothesis should be explored in future experiments.

We note that the present conclusions, which were based on comparisons between thresholds for 3D rotation and 3D translation stimuli several months (up to 2 yr) following LB, are not directly comparable to previous studies. We found persistent deficits in rotation detection thresholds, which were due to lower response gains, while response variance was unaltered. This result differs from the higher variability reported by Jamali et al. (2014), who tested yaw responses approximately 1 mo after LB. The present results show that response variability completely recovers in the long-term after LB (and when evaluated in 3D). On the other hand, we found no difference in translation detection thresholds or 3D gains for VN neurons in IN vs. LB animals. Thus deficits in gain and threshold reported by Newlands et al. (2014) and Liu et al. (2013a) appear to recover in steady state.

Finally, it is important to note that the relationship between the observed persistent increases in canal afferent variability and the recovered variability of central neurons chronically after LB is presently unknown. This is because of current limitations in understanding how input variability influences output variability. Such relationships would depend both on individual neuron properties, as well as on the correlated variability and information transfer among multiple neurons in the network. Future studies that simultaneously measure stimulus-driven responses across a large number of neurons, combined with theory, would be necessary to understand these relationships.

Next we briefly summarize the most relevant prior findings on vestibular afferent properties, vestibular compensation following unilateral labyrinth loss, as well as differences in the organization of the semicircular canals and otolith pathways. In addition, we also discuss the challenges associated with comparison of neuronal with behavioral thresholds when they were not measured simultaneously.

Basic Properties of Semicircular Canal Afferents and Efferents

Semicircular canal afferents were characterized by similar mean response properties as in previous studies (Haque et al. 2004; Ramachandran and Lisberger 2006; Sadeghi et al. 2007a, 2007b). The only difference between the present sample and previous reports is that we did not encounter many cells with high CV* and low response gains, which have been shown to correspond to canal afferents forming calyx endings in the central striola region of the chinchilla (Baird et al. 1988). Compared with dimorphic-irregular afferents, which are characterized by the highest gain across all canal afferents, calyx-irregular afferents have much lower gains than what would be expected by the typical monotonic relationship between response gain and CV* (Baird et al. 1988; Ramachandran and Lisberger 2006; Sadeghi et al. 2006). Such a group of irregular afferents (CV* > 0.35 and gain at 0.5 Hz < 0.5) was previously identified in the macaque by Ramachandran and Lisberger (2006), although other macaque studies also did not encounter them readily (Haque et al. 2004; Sadeghi et al. 2007a, 2007b). The reason for this difference is not known, but may reflect differences in recording methodologies; e.g., use of metal electrodes in the present study. Thus the conclusions reached here should be considered applicable to only regular and dimorphic irregular canal afferents.

The low yield of calyx-like canal afferents could also explain the two differences between the present IN results and those of Sadeghi et al. (2007b). First, we encountered a higher proportion of regular afferents (70–76% vs. 38–45%). Second, we found a small dependence of neural threshold on CV*, and this dependence was in the opposite direction to that reported by Sadeghi et al. (2007b). That irregular afferents are as sensitive (i.e., have lower threshold), or more sensitive, than regular afferents has also been reported for otolith afferents (Jamali et al. 2013; Yu et al. 2012) and is due to the fact that spike count variability is proportional to response gain (Fig. 7). Thus, for the majority of regular and irregular afferents, a strong positive correlation between response gain and spike count variance results in neuronal thresholds that show either no dependence on CV* (Jamali et al. 2013) or small improvement with increasing CV* (Fig. 5; Yu et al. 2012). This finding is common to both canal and otolith afferents. However, a separate population of low-gain putative calyx type canal afferents exists, and these fibers do not follow the same gain vs. variance relationship (Fig. 6, filled symbols). Thus the detection thresholds for putative calyx canal afferents may be the highest of all afferents due to low response gains and high spike count variance. In addition, it is possible that low-gain, calyx-type, irregular afferents might exhibit the largest increases in firing rate variability after LB (Fig. 5, filled red/black symbols).

Although we cannot exclude that the observed increase in spike count variability after LB is due to retrograde changes due to altered properties in VN neurons that receive inputs from both labyrinths, it is tempting to speculate that the observed changes in spike count variability might be mediated by the vestibular efferent system. Efferent fibers (from ∼300–400 brain stem cells in primates) make contacts presynaptically onto type II hair cells and postsynaptically onto afferent calyces (Klinke and Galley 1974; Lindeman 1969; Lysakowski and Goldberg 1997; Sans and Highstein 1984). Efferent-mediated effects on vestibular afferents are heterogeneous (Bernard et al. 1985; Boyle and Highstein 1990; Brichta and Goldberg 2000; Dickman and Correia 1993; Goldberg and Fernandez 1980; Highstein and Baker 1985; Holt et al. 2006; Marlinski et al. 2004; McCue and Guinan 1994; Rossi et al. 1980; Rossi and Martini 1991; Sugai et al. 1991; reviewed by Lysakowski and Goldberg 2004). Effects of efferent electrical stimulation are the largest in irregular afferents, consisting of both fast (10–100 ms time constant) and slow (5–20 s) components. In contrast, efferent effects on regular afferents are small and predominantly slow (Goldberg and Fernandez 1980; Marlinski et al. 2004; McCue and Guinan 1994).

Over the years, multiple functions have been suggested for the vestibular efferent system (see review by Goldberg 2000), and each has been systematically rejected. Vestibular afferent activity is similar during active vs. passive head movements and does not modulate during proprioceptive or visual stimulation (Cullen and Minor 2002; Miles and Braitman 1980). Similarly, the hypothesis that the vestibular efferent system could be used to balance activity between the two labyrinths and/or extend the dynamic range of the intact labyrinth after contralateral vestibular damage has also been refuted (Sadeghi et al. 2007b). Our results are completely consistent with these findings, showing no difference in mean response gain, phase and PD after labyrinthectomy.

VOR Deficits After LB

It has been well-established that loss of unilateral vestibular function results in both postural and VOR deficits, including a pronounced head tilt toward the lesion (Angelaki et al. 2000; Fetter and Zee 1988; Smith and Curthoys 1989). However, in contrast to incomplete passive VOR recovery, counter-rotation gains during combined eye-head gaze shifts (i.e., the compensatory eye counter-rotation that stabilizes gaze as the head movement continues during an eye-head gaze shift) recovered within 1 wk after LB, likely using extravestibular cues (Newlands et al. 2001).

Previous investigations have described two phases in compensation from unilateral loss (Curthoys and Halmagyi 1995; Fetter and Zee 1988; Lasker et al. 2000; Ris et al. 1995; Smith and Curthoys 1989). Within approximately the first week, recovery is fast and involves mainly compensation of the static imbalance and recovery of spontaneous firing rate in central neurons. VOR gain recovery is also fast during this period, likely due to the restoration of spontaneous activity in central pathways. The second compensation phase is slower and takes place within several weeks or months after unilateral loss. The goal of the present study was not to characterize the early deficits, but rather those after the process of compensation from vestibular imbalance had been completed.

The higher the peak acceleration of rotation, the smaller the recovery of the rVOR and the larger the asymmetry between responses to ipsilesional and contralesional rotation (Crane and Demer 1998; Gilchrist et al. 1998; Halmagyi et al. 1990; Sadeghi et al. 2006; Tabak et al. 1997; Vibert et al. 1993). In contrast, low acceleration rotations elicit a nearly complete recovery of the yaw rVOR in both humans and monkeys (Allum et al. 1988; Baloh et al. 1984; Fetter and Dichgans 1990; Fetter and Zee 1988; Takahashi et al. 1984; Takahashi and Igarashi 1977; Wolfe and Kos 1977). Because here we measured detection thresholds, stimuli were exclusively in the low magnitude range.

Changes in Neuronal Response Gain During Rotation

Commissural inhibition is fundamental to the central organization of semicircular canal signals (Kasahara and Uchino 1974; Precht et al. 1966; Precht and Shimazu 1965; Shimazu and Precht 1966). For example, VN neuron responses are driven by excitatory inputs from the ipsilateral horizontal semicircular canal and disinhibition from the contralateral horizontal canal. As a result of this push-pull arrangement, facilitation from the ipsilateral side and disinhibition from the contralateral side ensure higher gains in VN neurons than canal afferents. The significance of this bilateral organization was demonstrated long ago, by showing partial gain recovery after LB (Newlands and Perachio 1990a, 1990b; Ris and Godaux 1998; Smith and Curthoys 1988a, 1988b). More recent studies in alert macaques have shown small but persistent gain deficits for type I VN neurons postlesion (Newlands and Wei 2013; Sadeghi et al. 2011, 2010). Our results are in general agreement with these findings. We found that 3D PD gains after compensation remain lower than those in IN animals, even after several months postlesion.

Vestibular compensation involves multiple, parallel processes in plasticity (reviewed by Dutia 2010). Substantial changes in the intrinsic membrane properties (Beraneck et al. 2003, 2004; Cameron and Dutia 1997, 1999; Darlington et al. 2002; Him and Dutia 2001; Ris et al. 2002; Straka et al. 2005), synaptic reorganization of the brain stem vestibular pathways (Dieringer 2003; Goto et al. 2000, 2001, 2002; Johnston et al. 2001; Lim et al. 2010; Vibert et al. 2000; Yamanaka et al. 2000) and even neurogenesis (Dutheil et al. 2009) have all been documented. Vestibular compensation requires an intact vestibulo-cerebellum (Beraneck et al. 2008; Courjon et al. 1982; Haddad et al. 1977; Kim et al. 1997a, 1997b; Kitahara et al. 1997, 1998, 2000). Unilateral lesions of the vestibular end organs induce c-Fos expression throughout the vestibulo-olivo-cerebellar pathway in rodents (Cirelli et al. 1993, 1996; Kaufmann et al. 1992, 1993, 1999; Kim et al. 1997a, 1997b; Kitahara et al. 1997; Li et al. 2001; Sato et al. 1997; Shinder et al. 2005).

Changes in Neuronal Responses During Translation

Unlike the semicircular canals, the otolith receptors have a wide representation of PD of linear acceleration within a single macula, including opposite-directed responses across an imaginary line known as the striola (Flock 1964; Lindeman 1969). Indeed, in addition to the commissural system (Bai et al. 2002; Uchino et al. 1999, 2001), another mechanism, known as “cross-striolar inhibition,” was shown to exist and be stronger for the sacculus than the utricle (Ogawa et al. 2000; Uchino et al. 1997): VN neurons were excited monosynaptically by ipsilateral otolith afferents from one side of the striola and inhibited disynaptically by afferents from the other side of the striola of the same end organ. Thus appropriate push-pull afferent convergence in the VN could arise either from both labyrinths (as in the canal system), or from the same otolith macula through cross-striolar inhibition (Uchino et al. 1999).

Despite this redundancy, which could allow the otolith system to maintain the push-pull organization with just a single functional labyrinth, dramatic changes in the tVOR have been reported even after 3 mo of recovery following unilateral vestibular damage, particularly during transient displacements (Angelaki et al. 2000). Such asymmetric tVOR responses during ipsilesional and contralesional translation are very strong within the first week following lesion (Angelaki et al. 2000; Lempert et al. 1998; Tian et al. 2007). How much tVOR has recovered in chronic patients with unilaterally vestibular damage remains controversial (Bronstein et al. 1991; Tian et al. 2007). Asymmetric response gains have been traced down to VN neurons in anesthetized animals (Chan 1997; Chan et al. 1999; Hoshino and Pompeiano 1977; Lacour et al. 1985; Xerri et al. 1983). How signals from the remaining intact otolith organs reach the motoneurons through reorganization of ipsilesional and/or contralesional pathways (Angelaki et al. 2001; Green et al. 2001; Rohregger and Dieringer 2003) requires further study.

In summary, the present findings, showing complete recovery of translation response gains and detection thresholds, demonstrate that otolith afferent signals from a single labyrinth are sufficient after compensation to maintain the sensitivity and direction discriminability of linear acceleration responses in VN neurons. In contrast, recovery was incomplete during 3D rotations, where both response gains and rotation detection thresholds were lower than normal, particularly in the VN ipsilateral to the lesion.

It is important to highlight that direct comparisons with behavioral performance are not possible because behavioral thresholds have never been measured either in IN (but see Garcia-Johnston et al. 2013) or in LB macaques (but see Cousins et al. 2013; Cutfield et al. 2011; Grabherr et al. 2008; Haburcakova et al. 2012; Priesol et al. 2014; Valko et al. 2012 for human perceptual thresholds). We are only aware of Liu et al. (2013a) who showed persistent heading discrimination deficits for up to 3 mo post-LB. We note that, in the absence of simultaneous behavioral recordings, we have made no attempt to estimate population thresholds. This is because, as summarized next, we consider flawed any attempt to calculate population thresholds and pool size in the absence of simultaneously measured behavior because of many potentially erroneous assumptions.

How Can Neural Thresholds Be Compared with Behavioral Thresholds?

Several recent studies have measured neural thresholds in the vestibular periphery (canal afferents: Sadeghi et al. 2007a; otolith afferents: Jamali et al. 2013; Yu et al. 2012) and the vestibular brain stem (translation: Liu et al. 2010, 2013a, 2013b; rotation: Jamali et al. 2013, 2014; Massot et al. 2011). With the exception of Liu et al. (2013a, 2013b) in the VN (but see also Chen et al. 2013; Gu et al. 2007, 2011 for cortical areas), none of these studies actually measured neuronal thresholds simultaneously with behavioral thresholds. There are multiple reasons that direct comparisons between neural and behavioral thresholds should not be made unless measured simultaneously in the same animals using the same motion platform. First, behavioral thresholds vary from subject to subject (e.g., see Suppl. Fig. 2 of Liu et al. 2013b). Second, macaque and human thresholds are not necessarily similar; thus quantitative comparisons between macaque neurons and human behavior (as previously done by Massot et al. 2011 and Jamali et al. 2013) may lead to inappropriate conclusions. Third, the exact value measured for the neural threshold depends critically on the filtering and averaging used to compute trial-by-trial firing rates (Fig. 8; see also Fig. 9, Yu et al. 2012). Fourth, the inherent stimulus-driven vibrations in different motion systems might also result in different thresholds in different laboratories.

A few recent studies characterizing neuronal thresholds without simultaneously measuring perceptual thresholds have also computed population thresholds and characterized the extent of further convergence necessary to convert macaque neuronal thresholds into human perceptual thresholds (Jamali et al. 2013; Massot et al. 2011). In addition to the multiple problems associated with this approach summarized in the previous paragraph, an additional, serious caveat emerges when considering the fact that trial-by-trial firing rates show robust interneuronal correlations in the VN (Liu et al. 2013b). Such interneuronal correlations can affect both the information carried by the population, but also the decoding efficiency (see Liu et al. 2013a, 2013b for more details). Furthermore, when interneuronal correlations are present, population thresholds might not increase in proportion to the number of neurons in the pool, and, for the same neuronal pool size, different population thresholds will result, depending on whether the decoder has full knowledge of these correlations or not (Haefner et al. 2013; Pitkow et al. 2013). Furthermore, it is now well established that, to connect the dots between neural activity and perception, one must consider both 1) the decoder (optimal vs. suboptimal and there are multiple levels of suboptimality) and 2) interneuronal correlations. Some types of correlations can be removed with optimal decoding (by “reading” out the neural activity along the best axis, which always exists when correlations are in a different format from the signal). But other types of correlations (those which are identical to the signal) cannot be “removed” by any decoder, optimal or suboptimal (Beck et al. 2012). The only way to realistically tackle this challenge is by simultaneously measuring neural activity and perception. Recent theory has provided ways of inferring read-out weights, but one needs to know, not only neural thresholds, but also correlations among neurons and choice-probabilities (correlations between neuron and behavior) (Haefner et al. 2013).

We note that in the present study we have measured neural thresholds without simultaneously measuring perception. Thus we have made no attempt to compute population thresholds and to compare the present results with perception. Future studies employing simultaneously recorded neural and perceptual thresholds, which is an important requirement for quantitative simulations and predictions at the population level, are necessary to understand both the peripheral and central processing of vestibular signals.

GRANTS

This study was supported by National Institute on Deafness and Other Communications Disorders Grant DC-04260.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: X.-J.Y., J.S.T., and D.E.A. conception and design of research; X.-J.Y., J.S.T., and J.D.D. performed experiments; X.-J.Y., J.S.T., and D.E.A. analyzed data; X.-J.Y., J.S.T., and D.E.A. interpreted results of experiments; X.-J.Y. and J.S.T. prepared figures; X.-J.Y., J.S.T., J.D.D., and D.E.A. drafted manuscript; X.-J.Y., J.S.T., J.D.D., S.D.N., and D.E.A. edited and revised manuscript; X.-J.Y., J.S.T., J.D.D., S.D.N., and D.E.A. approved final version of manuscript.

REFERENCES

  1. Allum JH, Yamane M, Pfaltz CR. Long-term modifications of vertical and horizontal vestibulo-ocular reflex dynamics in man. I. After acute unilateral peripheral vestibular paralysis. Acta Otolaryngol (Stockh) 105: 328–337, 1988 [DOI] [PubMed] [Google Scholar]
  2. Angelaki DE. Dynamic polarization vector of spatially tuned neurons. IEEE Trans Biomed Eng 38: 1053–1060, 1991 [DOI] [PubMed] [Google Scholar]
  3. Angelaki DE. Spatio-temporal convergence (STC) in otolith neurons. Biol Cybern 67: 83–96, 1992 [DOI] [PubMed] [Google Scholar]
  4. Angelaki DE, Green AM, Dickman JD. Differential sensorimotor processing of vestibulo-ocular signals during rotation and translation. J Neurosci 21: 3968–3985, 2001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Angelaki DE, McHenry MQ, Dickman JD, Perachio AA. Primate translational vestibuloocular reflexes. III. Effects of bilateral labyrinthine electrical stimulation. J Neurophysiol 83: 1662–1676, 2000 [DOI] [PubMed] [Google Scholar]
  6. Bai R, Meng H, Sato H, Imagawa M, Sasaki M, Uchino Y. Properties of utricular-activated vestibular neurons that project to the contralateral vestibular nuclei in the cat. Exp Brain Res 147: 419–425, 2002 [DOI] [PubMed] [Google Scholar]
  7. Bair W, Zohary E, Newsome WT. Correlated firing in macaque visual area MT: time scales and relationship to behavior. J Neurosci 21: 1676–1697, 2001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Baird RA, Desmadryl G, Fernandez C, Goldberg JM. The vestibular nerve of the chinchilla. II. Relation between afferent response properties and peripheral innervation patterns in the semicircular canals. J Neurophysiol 60: 182–203, 1988 [DOI] [PubMed] [Google Scholar]
  9. Baloh RW, Hess K, Honrubia V, Yee RD. Low and high frequency sinusoidal rotational testing in patients with peripheral vestibular lesions. Acta Otolaryngol Suppl (Stockh) 406: 189–193, 1984 [DOI] [PubMed] [Google Scholar]
  10. Beck J, Moreno R, Latham P, Pouget A. Noise correlations in population codes with finite information (Abstract). COSYNE T-12: 2012 [Google Scholar]
  11. Beraneck M, Hachemaoui M, Idoux E, Ris L, Uno A, Godaux E, Vidal PP, Moore LE, Vibert N. Long-term plasticity of ipsilesional medial vestibular nucleus neurons after unilateral labyrinthectomy. J Neurophysiol 90: 184–203, 2003 [DOI] [PubMed] [Google Scholar]
  12. Beraneck M, Idoux E, Uno A, Vidal PP, Moore LE, Vibert N. Unilateral labyrinthectomy modifies the membrane properties of contralesional vestibular neurons. J Neurophysiol 92: 1668–1684, 2004 [DOI] [PubMed] [Google Scholar]
  13. Beraneck M, McKee JL, Aleisa M, Cullen KE. Asymmetric recovery in cerebellar-deficient mice following unilateral labyrinthectomy. J Neurophysiol 100: 945–958, 2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Bernard C, Cochran SL, Precht W. Presynaptic actions of cholinergic agents upon the hair cell-afferent fiber synapse in the vestibular labyrinth of the frog. Brain Res 338: 225–236, 1985 [DOI] [PubMed] [Google Scholar]
  15. Boyle R, Highstein SM. Efferent vestibular system in the toadfish: action upon horizontal semicircular canal afferents. J Neurosci 10: 1570–1582, 1990 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Brichta AM, Goldberg JM. Responses to efferent activation and excitatory response-intensity relations of turtle posterior-crista afferents. J Neurophysiol 83: 1224–1242, 2000 [DOI] [PubMed] [Google Scholar]
  17. Bronstein AM, Mossman S, Luxon LM. The neck-eye reflex in patients with reduced vestibular and optokinetic function. Brain 114: 1–11, 1991 [PubMed] [Google Scholar]
  18. Bronte-Stewart HM, Lisberger SG. Physiological properties of vestibular primary afferents that mediate motor learning and normal performance of the vestibulo-ocular reflex in monkeys. J Neurosci 14: 1290–1308, 1994 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Broussard DM, Bhatia JK, Jones GE. The dynamics of the vestibulo-ocular reflex after peripheral vestibular damage. I. Frequency-dependent asymmetry. Exp Brain Res 125: 353–364, 1999 [DOI] [PubMed] [Google Scholar]
  20. Cameron SA, Dutia MB. Cellular basis of vestibular compensation: changes in intrinsic excitability of MVN neurones. Neuroreport 8: 2595–2599, 1997 [DOI] [PubMed] [Google Scholar]
  21. Cameron SA, Dutia MB. Lesion-induced plasticity in rat vestibular nucleus neurones dependent on glucocorticoid receptor activation. J Physiol 518: 151–158, 1999 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Chan YS. The coding of head orientations in neurons of bilateral vestibular nuclei of cats after unilateral labyrinthectomy: response to off-vertical axis rotation. Exp Brain Res 114: 293–303, 1997 [DOI] [PubMed] [Google Scholar]
  23. Chan YS, Shum DK, Lai CH. Neuronal response sensitivity to bidirectional off-vertical axis rotations: a dimension of imbalance in the bilateral vestibular nuclei of cats after unilateral labyrinthectomy. Neuroscience 94: 831–843, 1999 [DOI] [PubMed] [Google Scholar]
  24. Chen A, Deangelis GC, Angelaki DE. Functional specializations of the ventral intraparietal area for multisensory heading discrimination. J Neurosci 33: 3567–3581, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Chen A, DeAngelis GC, Angelaki DE. Macaque parieto-insular vestibular cortex: responses to self-motion and optic flow. J Neurosci 30: 3022–3042, 2010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Cirelli C, Pompeiano M, D'Ascanio P, Arrighi P, Pompeiano O. c-fos Expression in the rat brain after unilateral labyrinthectomy and its relation to the uncompensated and compensated stages. Neuroscience 70: 515–546, 1996 [DOI] [PubMed] [Google Scholar]
  27. Cirelli C, Pompeiano M, D'Ascanio P, Pompeiano O. Early c-fos expression in the rat vestibular and olivocerebellar systems after unilateral labyrinthectomy. Arch Ital Biol 131: 71–74, 1993 [PubMed] [Google Scholar]
  28. Courjon JH, Flandrin JM, Jeannerod M, Schmid R. The role of the flocculus in vestibular compensation after hemilabyrinthectomy. Brain Res 239: 251–257, 1982 [DOI] [PubMed] [Google Scholar]
  29. Cousins S, Kaski D, Cutfield N, Seemungal B, Golding JF, Gresty M, Glasauer S, Bronstein AM. Vestibular perception following acute unilateral vestibular lesions. PLos One 8: e61862, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Crane BT, Demer JL. Human horizontal vestibulo-ocular reflex initiation: effects of acceleration, target distance, and unilateral deafferentation. J Neurophysiol 80: 1151–1166, 1998 [DOI] [PubMed] [Google Scholar]
  31. Cullen KE, Minor LB. Semicircular canal afferents similarly encode active and passive head-on-body rotations: implications for the role of vestibular efference. J Neurosci 22: RC226, 2002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Curthoys IS. Vestibular compensation and substitution. Curr Opin Neurol 13: 27–30, 2000 [DOI] [PubMed] [Google Scholar]
  33. Curthoys IS, Halmagyi GM. Vestibular compensation: a review of the oculomotor, neural, and clinical consequences of unilateral vestibular loss. J Vestib Res 5: 67–107, 1995 [PubMed] [Google Scholar]
  34. Cutfield NJ, Cousins S, Seemungal BM, Gresty MA, Bronstein AM. Vestibular perceptual thresholds to angular rotation in acute unilateral vestibular paresis and with galvanic stimulation. Ann N Y Acad Sci 1233: 256–262, 2011 [DOI] [PubMed] [Google Scholar]
  35. Darlington CL, Dutia MB, Smith PF. The contribution of the intrinsic excitability of vestibular nucleus neurons to recovery from vestibular damage. Eur J Neurosci 15: 1719–1727, 2002 [DOI] [PubMed] [Google Scholar]
  36. De Waele C, Graf W, Josset P, Vidal PP. A radiological analysis of the postural syndromes following hemilabyrinthectomy and selective canal and otolith lesions in the guinea pig. Exp Brain Res 77: 166–182, 1989 [DOI] [PubMed] [Google Scholar]
  37. Dickman JD, Angelaki DE. Three-dimensional organization of vestibular related eye movements during off-vertical axis rotation and linear translation in pigeons. Exp Brain Res 129: 391–400, 1999 [DOI] [PubMed] [Google Scholar]
  38. Dickman JD, Correia MJ. Bilateral communication between vestibular labyrinths in pigeons. Neuroscience 57: 1097–1108, 1993 [DOI] [PubMed] [Google Scholar]
  39. Dieringer N. Activity-related postlesional vestibular reorganization. Ann N Y Acad Sci 1004: 50–60, 2003 [PubMed] [Google Scholar]
  40. Dieringer N. “Vestibular compensation”: neural plasticity and its relations to functional recovery after labyrinthine lesions in frogs and other vertebrates. Prog Neurobiol 46: 97–129, 1995 [PubMed] [Google Scholar]
  41. Dutheil S, Brezun JM, Leonard J, Lacour M, Tighilet B. Neurogenesis and astrogenesis contribution to recovery of vestibular functions in the adult cat following unilateral vestibular neurectomy: cellular and behavioral evidence. Neuroscience 164: 1444–1456, 2009 [DOI] [PubMed] [Google Scholar]
  42. Dutia MB. Mechanisms of vestibular compensation: recent advances. Curr Opin Otolaryngol Head Neck Surg 18: 420–424, 2010 [DOI] [PubMed] [Google Scholar]
  43. Fernandez C, Goldberg JM. Physiology of peripheral neurons innervating semicircular canals of the squirrel monkey. II. Response to sinusoidal stimulation and dynamics of peripheral vestibular system. J Neurophysiol 34: 661–675, 1971 [DOI] [PubMed] [Google Scholar]
  44. Fetter M, Dichgans J. Adaptive mechanisms of VOR compensation after unilateral peripheral vestibular lesions in humans. J Vestib Res 1: 9–22, 1990 [PubMed] [Google Scholar]
  45. Fetter M, Zee DS. Recovery from unilateral labyrinthectomy in rhesus monkey. J Neurophysiol 59: 370–393, 1988 [DOI] [PubMed] [Google Scholar]
  46. Flock A. Structure of the macula utriculi with special reference to directional interplay of sensory responses as revealed by morphological polarization. J Cell Biol 22: 413–431, 1964 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Garcia-Johnston CD, Dickman JD, Angelaki DE. Vestibular behavioral thresholds for yaw rotation discrimination in rhesus macaques, Program No. 265.07. In: 2013 Neuroscience Meeting Planner San Diego, CA: Society for Neuroscience, 2013 [Google Scholar]
  48. Gilchrist DP, Curthoys IS, Cartwright AD, Burgess AM, Topple AN, Halmagyi M. High acceleration impulsive rotations reveal severe long-term deficits of the horizontal vestibulo-ocular reflex in the guinea pig. Exp Brain Res 123: 242–254, 1998 [DOI] [PubMed] [Google Scholar]
  49. Goldberg JM. Afferent diversity and the organization of central vestibular pathways. Exp Brain Res 130: 277–297, 2000 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Goldberg JM, Desmadryl G, Baird RA, Fernandez C. The vestibular nerve of the chinchilla. IV. Discharge properties of utricular afferents. J Neurophysiol 63: 781–790, 1990a [DOI] [PubMed] [Google Scholar]
  51. Goldberg JM, Desmadryl G, Baird RA, Fernandez C. The vestibular nerve of the chinchilla. V. Relation between afferent discharge properties and peripheral innervation patterns in the utricular macula. J Neurophysiol 63: 791–804, 1990b [DOI] [PubMed] [Google Scholar]
  52. Goldberg JM, Fernandez C. Efferent vestibular system in the squirrel monkey: anatomical location and influence on afferent activity. J Neurophysiol 43: 986–1025, 1980 [DOI] [PubMed] [Google Scholar]
  53. Goldberg JM, Fernandez C. Physiology of peripheral neurons innervating semicircular canals of the squirrel monkey. 3. Variations among units in their discharge properties. J Neurophysiol 34: 676–684, 1971a [DOI] [PubMed] [Google Scholar]
  54. Goldberg JM, Fernandez C. Physiology of peripheral neurons innervating semicircular canals of the squirrel monkey. I. Resting discharge and response to constant angular accelerations. J Neurophysiol 34: 635–660, 1971b [DOI] [PubMed] [Google Scholar]
  55. Goldberg JM, Smith CE, Fernandez C. Relation between discharge regularity and responses to externally applied galvanic currents in vestibular nerve afferents of the squirrel monkey. J Neurophysiol 51: 1236–1256, 1984 [DOI] [PubMed] [Google Scholar]
  56. Goto F, Straka H, Dieringer N. Expansion of afferent vestibular signals after the section of one of the vestibular nerve branches. J Neurophysiol 84: 581–584, 2000 [DOI] [PubMed] [Google Scholar]
  57. Goto F, Straka H, Dieringer N. Gradual and reversible central vestibular reorganization in frog after selective labyrinthine nerve branch lesions. Exp Brain Res 147: 374–386, 2002 [DOI] [PubMed] [Google Scholar]
  58. Goto F, Straka H, Dieringer N. Postlesional vestibular reorganization in frogs: evidence for a basic reaction pattern after nerve injury. J Neurophysiol 85: 2643–2646, 2001 [DOI] [PubMed] [Google Scholar]
  59. Grabherr L, Nicoucar K, Mast FW, Merfeld DM. Vestibular thresholds for yaw rotation about an earth-vertical axis as a function of frequency. Exp Brain Res 186: 677–681, 2008 [DOI] [PubMed] [Google Scholar]
  60. Green J, Christmas P, Goldmeier D, Byrne M, Kocsis A. A review of physical and psychological factors in vulvar vestibulitis syndrome. Int J STD AIDS 12: 705–709, 2001 [DOI] [PubMed] [Google Scholar]
  61. Gu Y, Angelaki DE, Deangelis GC. Neural correlates of multisensory cue integration in macaque MSTd. Nat Neurosci 11: 1201–1210, 2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Gu Y, DeAngelis GC, Angelaki DE. A functional link between area MSTd and heading perception based on vestibular signals. Nat Neurosci 10: 1038–1047, 2007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Gu Y, Liu S, Fetsch CR, Yang Y, Fok S, Sunkara A, DeAngelis GC, Angelaki DE. Perceptual learning reduces interneuronal correlations in macaque visual cortex. Neuron 71: 750–761, 2011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Gu Y, Watkins PV, Angelaki DE, DeAngelis GC. Visual and nonvisual contributions to three-dimensional heading selectivity in the medial superior temporal area. J Neurosci 26: 73–85, 2006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Haburcakova C, Lewis RF, Merfeld DM. Frequency dependence of vestibuloocular reflex thresholds. J Neurophysiol 107: 973–983, 2012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Haddad GM, Friendlich AR, Robinson DA. Compensation of nystagmus after VIIIth nerve lesions in vestibulo-cerebellectomized cats. Brain Res 135: 192–196, 1977 [DOI] [PubMed] [Google Scholar]
  67. Haefner RM, Gerwinn S, Macke JH, Bethge M. Inferring decoding strategies from choice probabilities in the presence of correlated variability. Nat Neurosci 16: 235–242, 2013 [DOI] [PubMed] [Google Scholar]
  68. Halmagyi GM, Curthoys IS, Cremer PD, Henderson CJ, Todd MJ, Staples MJ, D'Cruz DM. The human horizontal vestibulo-ocular reflex in response to high-acceleration stimulation before and after unilateral vestibular neurectomy. Exp Brain Res 81: 479–490, 1990 [DOI] [PubMed] [Google Scholar]
  69. Haque A, Angelaki DE, Dickman JD. Spatial tuning and dynamics of vestibular semicircular canal afferents in rhesus monkeys. Exp Brain Res 155: 81–90, 2004 [DOI] [PubMed] [Google Scholar]
  70. Highstein SM, Baker R. Action of the efferent vestibular system on primary afferents in the toadfish, Opsanus tau. J Neurophysiol 54: 370–384, 1985 [DOI] [PubMed] [Google Scholar]
  71. Him A, Dutia MB. Intrinsic excitability changes in vestibular nucleus neurons after unilateral deafferentation. Brain Res 908: 58–66, 2001 [DOI] [PubMed] [Google Scholar]
  72. Holt JC, Lysakowski A, Goldberg JM. Mechanisms of efferent-mediated responses in the turtle posterior crista. J Neurosci 26: 13180–13193, 2006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Hoshino K, Pompeiano O. Crossed responses of lateral vestibular neurons to macular labyrinthine stimulation. Brain Res 131: 152–157, 1977 [DOI] [PubMed] [Google Scholar]
  74. Jamali M, Carriot J, Chacron MJ, Cullen KE. Strong correlations between sensitivity and variability give rise to constant discrimination thresholds across the otolith afferent population. J Neurosci 33: 11302–11313, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Jamali M, Mitchell DE, Dale A, Carriot J, Sadeghi SG, Cullen KE. Neuronal detection thresholds during vestibular compensation: contributions of response variability and sensory substitution. J Physiol 592: 1565–1580, 2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  76. Johnston AR, Him A, Dutia MB. Differential regulation of GABA(A) and GABA(B) receptors during vestibular compensation. Neuroreport 12: 597–600, 2001 [DOI] [PubMed] [Google Scholar]
  77. Kasahara M, Uchino Y. Bilateral semicircular canal inputs to neurons in cat vestibular nuclei. Exp Brain Res 20: 285–296, 1974 [DOI] [PubMed] [Google Scholar]
  78. Kaufman GD, Anderson JH, Beitz AJ. Brainstem fos expression following acute unilateral labyrinthectomy in the rat. Neuroreport 3: 829–832, 1992 [DOI] [PubMed] [Google Scholar]
  79. Kaufman GD, Anderson JH, Beitz AJ. Otolith-brain stem connectivity: evidence for differential neural activation by vestibular hair cells based on quantification of fos expression in unilateral labyrinthectomized rats. J Neurophysiol 70: 117–127, 1993 [DOI] [PubMed] [Google Scholar]
  80. Kaufman GD, Shinder ME, Perachio AA. Correlation of fos expression and circling asymmetry during gerbil vestibular compensation. Brain Res 817: 246–255, 1999 [DOI] [PubMed] [Google Scholar]
  81. Kim MS, Jin BK, Chun SW, Lee MY, Lee SH, Kim JH, Park BR. Effect of MK801 on cFos-like protein expression in the medial vestibular nucleus at early stage of vestibular compensation in uvulonodullectomized rats. Neurosci Lett 231: 147–150, 1997a [DOI] [PubMed] [Google Scholar]
  82. Kim MS, Jin BK, Chun SW, Lee MY, Lee SH, Kim JH, Park BR. Role of vestibulocerebellar N-methyl-d-aspartate receptors for behavioral recovery following unilateral labyrinthectomy in rats. Neurosci Lett 222: 171–174, 1997b [DOI] [PubMed] [Google Scholar]
  83. Kitahara T, Fukushima M, Takeda N, Saika T, Kubo T. Effects of preflocculectomy on Fos expression and NMDA receptor-mediated neural circuits in the central vestibular system after unilateral labyrinthectomy. Acta Otolaryngol (Stockh) 120: 866–871, 2000 [DOI] [PubMed] [Google Scholar]
  84. Kitahara T, Takeda N, Saika T, Kubo T, Kiyama H. Role of the flocculus in the development of vestibular compensation: immunohistochemical studies with retrograde tracing and flocculectomy using Fos expression as a marker in the rat brainstem. Neuroscience 76: 571–580, 1997 [DOI] [PubMed] [Google Scholar]
  85. Kitahara T, Takeda N, Uno A, Kubo T, Mishina M, Kiyama H. Unilateral labyrinthectomy downregulates glutamate receptor delta-2 expression in the rat vestibulocerebellum. Brain Res Mol Brain Res 61: 170–178, 1998 [DOI] [PubMed] [Google Scholar]
  86. Klier EM, Meng H, Angelaki DE. Three-dimensional kinematics at the level of the oculomotor plant. J Neurosci 26: 2732–2737, 2006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  87. Klinke R, Galley N. Efferent innervation of vestibular and auditory receptors. Physiol Rev 54: 316–357, 1974 [DOI] [PubMed] [Google Scholar]
  88. Lacour M, Manzoni D, Pompeiano O, Xerri C. Central compensation of vestibular deficits. III. Response characteristics of lateral vestibular neurons to roll tilt after contralateral labyrinth deafferentation. J Neurophysiol 54: 988–1005, 1985 [DOI] [PubMed] [Google Scholar]
  89. Lasker DM, Hullar TE, Minor LB. Horizontal vestibuloocular reflex evoked by high-acceleration rotations in the squirrel monkey. III. Responses after labyrinthectomy. J Neurophysiol 83: 2482–2496, 2000 [DOI] [PubMed] [Google Scholar]
  90. Lempert T, Gianna C, Brookes G, Bronstein A, Gresty M. Horizontal otolith-ocular responses in humans after unilateral vestibular deafferentation. Exp Brain Res 118: 533–540, 1998 [DOI] [PubMed] [Google Scholar]
  91. Li YX, Tokuyama W, Okuno H, Miyashita Y, Hashimoto T. Differential induction of brain-derived neurotrophic factor mRNA in rat inferior olive subregions following unilateral labyrinthectomy. Neuroscience 106: 385–394, 2001 [DOI] [PubMed] [Google Scholar]
  92. Lim R, Callister RJ, Brichta AM. An increase in glycinergic quantal amplitude and frequency during early vestibular compensation in mouse. J Neurophysiol 103: 16–24, 2010 [DOI] [PubMed] [Google Scholar]
  93. Lindeman HH. Studies on the morphology of the sensory regions of the vestibular apparatus with 45 figures. Ergeb Anat Entwicklungsgesch 42: 1–113, 1969 [PubMed] [Google Scholar]
  94. Liu S, Dickman JD, Newlands SD, DeAngelis GC, Angelaki DE. Reduced choice-related activity and correlated noise accompany perceptual deficits following unilateral vestibular lesion. Proc Natl Acad Sci U S A 110: 17999–18004, 2013a [DOI] [PMC free article] [PubMed] [Google Scholar]
  95. Liu S, Gu Y, DeAngelis GC, Angelaki DE. Choice-related activity and correlated noise in subcortical vestibular neurons. Nat Neurosci 16: 89–97, 2013b [DOI] [PMC free article] [PubMed] [Google Scholar]
  96. Liu S, Yakusheva T, Deangelis GC, Angelaki DE. Direction discrimination thresholds of vestibular and cerebellar nuclei neurons. J Neurosci 30: 439–448, 2010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  97. Lysakowski A, Goldberg JM. Morphophysiology of the vestibular periphery. In: The Vestibular System, edited by Highstein SM, Popper A, Fay RR. New York: Springer-Verlag, 2004, p. 57–152 [Google Scholar]
  98. Lysakowski A, Goldberg JM. A regional ultrastructural analysis of the cellular and synaptic architecture in the chinchilla cristae ampullares. J Comp Neurol 389: 419–443, 1997 [DOI] [PMC free article] [PubMed] [Google Scholar]
  99. Marlinski V, McCrea RA. Coding of self-motion signals in ventro-posterior thalamus neurons in the alert squirrel monkey. Exp Brain Res 189: 463–472, 2008 [DOI] [PubMed] [Google Scholar]
  100. Marlinski V, Plotnik M, Goldberg JM. Efferent actions in the chinchilla vestibular labyrinth. J Assoc Res Otolaryngol 5: 126–143, 2004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  101. Massot C, Chacron MJ, Cullen KE. Information transmission and detection thresholds in the vestibular nuclei: single neurons vs. population encoding. J Neurophysiol 105: 1798–1814, 2011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  102. McCue MP, Guinan JJ., Jr Influence of efferent stimulation on acoustically responsive vestibular afferents in the cat. J Neurosci 14: 6071–6083, 1994 [DOI] [PMC free article] [PubMed] [Google Scholar]
  103. Meng H, Green AM, Dickman JD, Angelaki DE. Pursuit–vestibular interactions in brain stem neurons during rotation and translation. J Neurophysiol 93: 3418–3433, 2005 [DOI] [PubMed] [Google Scholar]
  104. Meng H, May PJ, Dickman JD, Angelaki DE. Vestibular signals in primate thalamus: properties and origins. J Neurosci 27: 13590–13602, 2007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  105. Miles FA, Braitman DJ. Long-term adaptive changes in primate vestibuloocular reflex. II. Electrophysiological observations on semicircular canal primary afferents. J Neurophysiol 43: 1426–1436, 1980 [DOI] [PubMed] [Google Scholar]
  106. Newlands SD, Lin N, Wei M. Responses of non-eye movement central vestibular neurons to sinusoidal horizontal translation in compensated macaques after unilateral labyrinthectomy. J Neurophysiol. In press, 2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  107. Newlands SD, Dara S, Kaufman GD. Relationship of static and dynamic mechanisms in vestibuloocular reflex compensation. Laryngoscope 115: 191–204, 2005 [DOI] [PubMed] [Google Scholar]
  108. Newlands SD, Hesse SV, Haque A, Angelaki DE. Head unrestrained horizontal gaze shifts after unilateral labyrinthectomy in the rhesus monkey. Exp Brain Res 140: 25–33, 2001 [DOI] [PubMed] [Google Scholar]
  109. Newlands SD, Perachio AA. Compensation of horizontal canal related activity in the medial vestibular nucleus following unilateral labyrinth ablation in the decerebrate gerbil. I. Type I neurons. Exp Brain Res 82: 359–372, 1990a [DOI] [PubMed] [Google Scholar]
  110. Newlands SD, Perachio AA. Compensation of horizontal canal related activity in the medial vestibular nucleus following unilateral labyrinth ablation in the decerebrate gerbil. II. Type II neurons. Exp Brain Res 82: 373–383, 1990b [DOI] [PubMed] [Google Scholar]
  111. Newlands SD, Wei M. Responses of central vestibular neurons to sinusoidal yaw rotation in compensated macaques after unilateral labyrinthectomy. J Neurophysiol 110: 1822–1836, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  112. Ogawa Y, Kushiro K, Zakir M, Sato H, Uchino Y. Neuronal organization of the utricular macula concerned with innervation of single vestibular neurons in the cat. Neurosci Lett 278: 89–92, 2000 [DOI] [PubMed] [Google Scholar]
  113. Paige GD. Vestibuloocular reflex and its interactions with visual following mechanisms in the squirrel monkey. II. Response characteristics and plasticity following unilateral inactivation of horizontal canal. J Neurophysiol 49: 152–168, 1983 [DOI] [PubMed] [Google Scholar]
  114. Pitkow XS, Liu S, Angelaki DE, Deangelis GC, Pouget A. How can single sensory neurons predict behavior? (Abstract). Soc Neurosci 43: 15.–02., 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  115. Precht W, Shimazu H. Functional connections of tonic and kinetic vestibular neurons with primary vestibular afferents. J Neurophysiol 28: 1014–1028, 1965 [DOI] [PubMed] [Google Scholar]
  116. Precht W, Shimazu H, Markham CH. A mechanism of central compensation of vestibular function following hemilabyrinthectomy. J Neurophysiol 29: 996–1010, 1966 [DOI] [PubMed] [Google Scholar]
  117. Priesol AJ, Valko Y, Merfeld DM, Lewis RF. Motion perception in patients with idiopathic bilateral vestibular hypofunction. Otolaryngol Head Neck Surg 150: 1040–1042, 2014 [DOI] [PubMed] [Google Scholar]
  118. Ramachandran R, Lisberger SG. Transformation of vestibular signals into motor commands in the vestibuloocular reflex pathways of monkeys. J Neurophysiol 96: 1061–1074, 2006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  119. Ris L, de Waele C, Serafin M, Vidal PP, Godaux E. Neuronal activity in the ipsilateral vestibular nucleus following unilateral labyrinthectomy in the alert guinea pig. J Neurophysiol 74: 2087–2099, 1995 [DOI] [PubMed] [Google Scholar]
  120. Ris L, Godaux E. Neuronal activity in the vestibular nuclei after contralateral or bilateral labyrinthectomy in the alert guinea pig. J Neurophysiol 80: 2352–2367, 1998 [DOI] [PubMed] [Google Scholar]
  121. Ris L, Hachemaoui M, Godaux E. Effect of labyrinthectomy on the spike generator of vestibular neurons in the guinea pig. Neuroreport 13: 1875–1879, 2002 [DOI] [PubMed] [Google Scholar]
  122. Rohregger M, Dieringer N. Postlesional vestibular reorganization improves the gain but impairs the spatial tuning of the maculo-ocular reflex in frogs. J Neurophysiol 90: 3736–3749, 2003 [DOI] [PubMed] [Google Scholar]
  123. Rossi ML, Martini M. Efferent control of posterior canal afferent receptor discharge in the frog labyrinth. Brain Res 555: 123–134, 1991 [DOI] [PubMed] [Google Scholar]
  124. Rossi ML, Prigioni I, Valli P, Casella C. Activation of the efferent system in the isolated frog labyrinth: effects on the afferent EPSPs and spike discharge recorded from single fibres of the posterior nerve. Brain Res 185: 125–137, 1980 [DOI] [PubMed] [Google Scholar]
  125. Sadeghi SG, Chacron MJ, Taylor MC, Cullen KE. Neural variability, detection thresholds, and information transmission in the vestibular system. J Neurosci 27: 771–781, 2007a [DOI] [PMC free article] [PubMed] [Google Scholar]
  126. Sadeghi SG, Minor LB, Cullen KE. Dynamics of the horizontal vestibuloocular reflex after unilateral labyrinthectomy: response to high frequency, high acceleration, and high velocity rotations. Exp Brain Res 175: 471–484, 2006 [DOI] [PubMed] [Google Scholar]
  127. Sadeghi SG, Minor LB, Cullen KE. Multimodal integration after unilateral labyrinthine lesion: single vestibular nuclei neuron responses and implications for postural compensation. J Neurophysiol 105: 661–673, 2011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  128. Sadeghi SG, Minor LB, Cullen KE. Neural correlates of motor learning in the vestibulo-ocular reflex: dynamic regulation of multimodal integration in the macaque vestibular system. J Neurosci 30: 10158–10168, 2010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  129. Sadeghi SG, Minor LB, Cullen KE. Response of vestibular-nerve afferents to active and passive rotations under normal conditions and after unilateral labyrinthectomy. J Neurophysiol 97: 1503–1514, 2007b [DOI] [PubMed] [Google Scholar]
  130. Sans A, Highstein SM. New ultrastructural features in the vestibular labyrinth of the toadfish, Opsanus tau. Brain Res 308: 191–195, 1984 [DOI] [PubMed] [Google Scholar]
  131. Sato T, Tokuyama W, Miyashita Y, Okuno H. Temporal and spatial dissociation of expression patterns between Zif268 and c-Fos in rat inferior olive during vestibular compensation. Neuroreport 8: 1891–1895, 1997 [DOI] [PubMed] [Google Scholar]
  132. Schor RH, Angelaki DE. The algebra of neural response vectors. Ann N Y Acad Sci 656: 190–204, 1992 [DOI] [PubMed] [Google Scholar]
  133. Shimazu H, Precht W. Inhibition of central vestibular neurons from the contralateral labyrinth and its mediating pathway. J Neurophysiol 29: 467–492, 1966 [DOI] [PubMed] [Google Scholar]
  134. Shinder ME, Perachio AA, Kaufman GD. VOR and Fos response during acute vestibular compensation in the Mongolian gerbil in darkness and in light. Brain Res 1038: 183–197, 2005 [DOI] [PubMed] [Google Scholar]
  135. Sirkin DW, Precht W, Courjon JH. Initial, rapid phase of recovery from unilateral vestibular lesion in rat not dependent on survival of central portion of vestibular nerve. Brain Res 302: 245–256, 1984 [DOI] [PubMed] [Google Scholar]
  136. Smith PF, Curthoys IS. Mechanisms of recovery following unilateral labyrinthectomy: a review. Brain Res Brain Res Rev 14: 155–180, 1989 [DOI] [PubMed] [Google Scholar]
  137. Smith PF, Curthoys IS. Neuronal activity in the contralateral medial vestibular nucleus of the guinea pig following unilateral labyrinthectomy. Brain Res 444: 295–307, 1988a [DOI] [PubMed] [Google Scholar]
  138. Smith PF, Curthoys IS. Neuronal activity in the ipsilateral medial vestibular nucleus of the guinea pig following unilateral labyrinthectomy. Brain Res 444: 308–319, 1988b [DOI] [PubMed] [Google Scholar]
  139. Straka H, Vibert N, Vidal PP, Moore LE, Dutia MB. Intrinsic membrane properties of vertebrate vestibular neurons: function, development and plasticity. Prog Neurobiol 76: 349–392, 2005 [DOI] [PubMed] [Google Scholar]
  140. Sugai T, Sugitani M, Ooyama H. Effects of activation of the divergent efferent fibers on the spontaneous activity of vestibular afferent fibers in the toad. Jpn J Physiol 41: 217–232, 1991 [DOI] [PubMed] [Google Scholar]
  141. Tabak S, Collewijn H, Boumans LJ. Deviation of the subjective vertical in long-standing unilateral vestibular loss. Acta Otolaryngol (Stockh) 117: 1–6, 1997 [DOI] [PubMed] [Google Scholar]
  142. Takahashi M, Igarashi M, Homick JL. Effect of otolith end organ ablation on horizontal optokinetic nystagmus, and optokinetic afternystagmus in the squirrel monkey. ORL J Otorhinolaryngol Relat Spec 39: 74–81, 1977 [DOI] [PubMed] [Google Scholar]
  143. Takahashi M, Uemura T, Fujishiro T. Recovery of vestibulo-ocular reflex and gaze disturbance in patients with unilateral loss of labyrinthine function. Ann Otol Rhinol Laryngol 93: 170–175, 1984 [DOI] [PubMed] [Google Scholar]
  144. Takahashi K, Gu Y, May PJ, Newlands SD, DeAngelis GC, Angelaki DE. Multimodal coding of three-dimensional rotation and translation in area MSTd: comparison of visual and vestibular selectivity. J Neurosci 27: 9742–9756, 2007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  145. Takahashi M, Igarashi M. Comparison of vertical and horizontal optokinetic nystagmus in the squirrel monkey. ORL J Otorhinolaryngol Relat Spec 39: 321–329, 1977 [DOI] [PubMed] [Google Scholar]
  146. Tian JR, Ishiyama A, Demer JL. Temporal dynamics of semicircular canal and otolith function following acute unilateral vestibular deafferentation in humans. Exp Brain Res 178: 529–541, 2007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  147. Uchino Y, Kushiro K. Differences between otolith- and semicircular canal-activated neural circuitry in the vestibular system. Neurosci Res 71: 315–327, 2011 [DOI] [PubMed] [Google Scholar]
  148. Uchino Y, Sasaki M, Sato H, Imagawa M, Suwa H, Isu N. Utricular input to cat extraocular motoneurons. Acta Otolaryngol Suppl (Stockh) 528: 44–48, 1997 [PubMed] [Google Scholar]
  149. Uchino Y, Sato H, Kushiro K, Zakir M, Imagawa M, Ogawa Y, Katsuta M, Isu N. Cross-striolar and commissural inhibition in the otolith system. Ann N Y Acad Sci 871: 162–172, 1999 [DOI] [PubMed] [Google Scholar]
  150. Uchino Y, Sato H, Zakir M, Kushiro K, Imagawa M, Ogawa Y, Ono S, Meng H, Zhang X, Katsuta M, Isu N, Wilson VJ. Commissural effects in the otolith system. Exp Brain Res 136: 421–430, 2001 [DOI] [PubMed] [Google Scholar]
  151. Valko Y, Lewis RF, Priesol AJ, Merfeld DM. Vestibular labyrinth contributions to human whole-body motion discrimination. J Neurosci 32: 13537–13542, 2012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  152. Vibert D, Safran AB, Hausler R. [Clinical evaluation of otolithic function by the measurement of ocular cyclotorsion and skew deviation]. Ann Otolaryngol Chir Cervicofac 110: 87–91, 1993 [PubMed] [Google Scholar]
  153. Vibert N, Beraneck M, Bantikyan A, Vidal PP. Vestibular compensation modifies the sensitivity of vestibular neurones to inhibitory amino acids. Neuroreport 11: 1921–1927, 2000 [DOI] [PubMed] [Google Scholar]
  154. Vibert N, De Waele C, Serafin M, Babalian A, Muhlethaler M, Vidal PP. The vestibular system as a model of sensorimotor transformations. A combined in vivo and in vitro approach to study the cellular mechanisms of gaze and posture stabilization in mammals. Prog Neurobiol 51: 243–286, 1997 [DOI] [PubMed] [Google Scholar]
  155. Wolfe JW, Kos CM. Nystagmic responses of the rhesus monkey to rotational stimulation following unilateral labyrinthectomy: final report. Trans Sect Otolaryngol Am Acad Ophthalmol Otolaryngol 84: 38–45, 1977 [PubMed] [Google Scholar]
  156. Xerri C, Gianni S, Manzoni D, Pompeiano O. Central compensation of vestibular deficits. I. Response characteristics of lateral vestibular neurons to roll tilt after ipsilateral labyrinth deafferentation. J Neurophysiol 50: 428–448, 1983 [DOI] [PubMed] [Google Scholar]
  157. Yamanaka T, Him A, Cameron SA, Dutia MB. Rapid compensatory changes in GABA receptor efficacy in rat vestibular neurones after unilateral labyrinthectomy. J Physiol 523: 413–424, 2000 [DOI] [PMC free article] [PubMed] [Google Scholar]
  158. Yu XJ, Dickman JD, Angelaki DE. Detection thresholds of macaque otolith afferents. J Neurosci 32: 8306–8316, 2012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  159. Zohary E, Shadlen MN, Newsome WT. Correlated neuronal discharge rate and its implications for psychophysical performance. Nature 370: 140–143, 1994 [DOI] [PubMed] [Google Scholar]

Articles from Journal of Neurophysiology are provided here courtesy of American Physiological Society

RESOURCES