INTRODUCTION

The signals carried by neurons in the vestibular system provide an important source of cognitive perception of self-motion in space. The signals that originate at the vestibular end organs are conveyed to the brain stem vestibular nuclei. From there, they are forwarded to the thalamus and are then sent to the cerebral cortex. The processing of these signals in the vestibular nuclei and various areas of the cerebral cortex, which are involved in perception of spatial motion, has been studied in great detail (Bremmer et al. 2002; Cullen and Roy 2004; Dickman and Angelaki 2004; Gdowski and McCrea 1999; Grüsser et al. 1990; Gu et al. 2006; Khalsa et al. 1987; Klam and Graf 2003, 2006; McCrea et al. 1999; Roy and Cullen 2001; Yakushin et al. 2006). However, thalamic neuronal activity associated with these signals has been examined less extensively and additional study is necessary to understand how motion signals are processed by the thalamic neurons that mediate passage of vestibular information to the cerebral cortex.

The focus of the present study is the ventroposterior thalamus of the squirrel monkey, which contains neurons sensitive to spatial motion. In contrast to thalamic relay nuclei for other sensory modalities, no distinct nucleus associated with transmission of vestibular sensory signals has been identified. Instead, vestibular-related neurons appear to be distributed throughout the broad ventroposterior portion of the thalamus (Büttner and Henn 1976; Büttner et al. 1977; Lang et al. 1979; Liedgren et al. 1976; Magnin and Fuchs 1977; Meng et al. 2007). These scattered neurons form projections to several cortical areas involved in processing self-motion signals including the parietal insular, frontoparietal, medial temporal, and posterior parietal cortices (Akbarian et al. 1992; Bremmer et al. 2002; Gu et al. 2006; Guldin and Grüsser 1998; Klam and Graf 2003, 2006).

Previous studies in rhesus monkeys found that thalamic neurons sensitive to vestibular stimulation had low spontaneous discharge rate and irregular firing pattern when compared with brain stem and cerebellar neurons sensitive to vestibular stimulation (Büttner and Henn 1976; Büttner et al. 1977; Magnin and Fuchs 1977). The responses of thalamic neurons to passive whole body rotation in the horizontal plane have been reported to lead in phase the angular velocity (Büttner et al. 1977; Meng et al. 2007).

In this paper we present our observations on ventroposterior thalamus neurons sensitive to whole animal rotation and/or translation in the horizontal plane. We have found these neurons to be clustered primarily on the borders of the ventroposterior nuclei. The rotational sensitivity of the thalamic neurons is inversely proportional to stimulus velocity across the range 4–100°/s. The translational sensitivity of these neurons decreases with increasing acceleration across the range of 0. 05–0.2 g, with the sensitivity inversely proportional to the square root of stimulus acceleration. Preliminary reports have been published elsewhere (Marlinski et al. 2004a,b).

METHODS

Experiments were carried out on two alert squirrel monkeys (Saimiri sciureus) prepared for chronic recording from single neurons. Protocols were approved by the University of Chicago Institutional Animal Care and Use Committee and were in compliance with the National Institutes of Health guidelines for the care and use of animals in research.

Surgical preparation

Surgical procedures were carried out under sterile conditions in a central surgical suite with the assistance of local veterinarians. Anesthesia was initiated with ketamine [10 mg/kg, administered intramuscularly (im)] and maintained with intravenous injection of remifentanil (0.025–0.075 μg·kg⁻¹·min⁻¹) and isoflurane inhalation (∼1%). The animals were supported postoperatively with bupenorphine (0.03 mg/kg, im) and sulfatrimethoprim (30 mg/kg, im).

Details of the surgical preparation have been described previously (Gdowski and McCrea 1999; McCrea et al. 1999). Surgical preparation included cranial attachment of a head-holding acrylic frame oriented parallel to the cranial horizontal (Frankfurt) plane (Emmers and Akert 1963). The frame featured an aluminum ring, which allowed fine adjustment of head position, and also served as a support for an electrode microadvance device. A Teflon-coated coil of stainless steel wire (Cooner) was implanted in one eye for the recording eye movements using the magnetic search coil technique (Robinson 1963). Labyrinthine stimulating electrodes of Teflon-coated silver wire (Medwire) were implanted bilaterally in the middle ear (Minor and Goldberg 1991).

Experimental setup

The monkey was seated on a perch situated on a revolving platform atop an actuator for rotational and translational movements. Mounted on the platform were structures used for restraint of the animal's head and body, electromagnetic field coils for eye and head movement measurements, and a three-dimensional manipulator for orienting a recording electrode microdrive.

The revolving platform was mounted on a linear sled (T3D, Trilogy Systems) with the translational axis in the horizontal plane. The platform supporting the monkey could be rotated around the yaw axis and locked at different angles relative to the translational axis. The linear sled was mounted on the top of rotary position servomotor (Kollmorgen Goldline DDR), the axis of which was aligned with the earth-vertical axis for producing yaw rotations. Motion profile commands were constructed using Spike2 (CED 1401) software, which controlled inputs to the servomotors.

During experiments, movements of the monkey's body were restrained by a jacket tethered to table structures, whereas limbs were free to grasp hand and foot rails. Care was taken to keep the animal in a comfortable posture similar to that observed during voluntary seating in its cage. The head was restrained by attaching the chronically implanted aluminum ring to a vertical rod fixed to the table structures.

At the beginning of recording from the thalamus the lateral geniculate body (LG) served as a topographical landmark in the area of investigation. Location of the LG was estimated according to field potentials evoked by 1-ms flash illumination (Vivitar 2000) of the darkened recording room.

Experimental protocol

Vestibular stimulation was achieved with the animal's rotations around the earth-vertical axis and translations in the horizontal plane along nasooccipital and interaural axes. The rotational vestibular stimuli were 0.25- to 2.0-Hz sinusoids of 4–100°/s peak angular velocity and the translational vestibular stimuli were 0.5- to 4.0-Hz sinusoids of 0.05–0.2 g (0.491–1.944 m/s²) peak acceleration.

Single-unit recording

Neuronal activity was recorded extracellularly with 4- to 7-MΩ epoxy-insulated tungsten microelectrodes (A-M Systems). Penetrations were carried out with the help of a stereotaxically positioned guide tube (25 gauge). The microelectrode was advanced into the thalamus using a micropositioner (FHC) secured to the head-holding frame during experiments.

After two-stage amplification and 10²- to 10⁴-Hz band-pass filtering of the microelectrode recording, the spikes were discriminated with a dual-window discriminator (Bak). Trigger pulses generated by the discriminator were sent to the event channel of the CED 1401 device.

Identification of neuronal inputs

The area of the thalamus that we explored in this study consisted predominantly of the ventroposterior lateral and medial nuclei related to the processing of somatosensory signals. Spatial motion of the animal triggers neuronal responses to vestibular stimulation, as well as neuronal responses due to somatosensory and visual stimuli. Our task was to isolate the neuronal responses related to the vestibular system from neuronal responses due to the other stimuli.

Somatosensory sensitivity of every neuron was examined by measuring its response to gentle touch of the hair and skin and application of pressure to the head, trunk, extremities, and the tail. When a neuron responded to any of these stimuli, the somatosensory receptive field was localized. Care was taken to avoid mechanical stimulation of this field during whole animal rotation or translation. If mechanical stimulation of such a field could not be excluded because of the animal's body motion relative to the harness during vestibular stimulation, the neuron was considered sensitive to somatosensory input only. Several neurons sensitive to the animal's motion responded to stimuli applied to a somatosensory receptive field that was not in contact with the harness. We considered those neurons sensitive to both vestibular and somatosensory inputs.

Neuronal sensitivity to neck proprioception was tested with rotations of the trunk while the head was held stationary. We recorded from groups of neurons that were responsive to both whole body and trunk rotations and thus had receiving vestibular and neck proprioceptive inputs. These neurons will be characterized in detail in a subsequent report.

To identify neurons receiving vestibular input, the whole animal was manually rotated and translated in the illuminated room. During those motions the animal faced a wall covered with a pattern of alternating black–white undulating vertical stripes with randomly varying 2- to 4-cm width. If neuronal discharge rate was altered during such motion the illumination was switched off and the neuron's sensitivity to motion was retested in the dark. If the neuron did not respond during motion in darkness, it was considered sensitive to visual input only. If the neuron was responsive to motion in darkness, it was regarded as a neuron receiving vestibular input.

Data acquisition

Eye and head movements were monitored with a search coil system (Angle-Meter NT, Primelec). Platform rotational velocity was recorded with an angular velocity sensor (Watson). Linear translations along two orthogonal axes were recorded with the dual-axis accelerometer (Kionix).

Analogue signals were low-pass filtered with six-pole 7-kHz Butterworth filters (Frequency Devices), digitized at 992 Hz with CED Power 1401, and saved on a computer for off-line analysis. In addition, unfiltered microelectrode recordings were sampled at the rate of 55.6 kHz and were used for field potential analysis and checking the fidelity of spike discrimination.

Data analysis

Data analysis was carried out using routines written for the Igor Pro (WaveMetrics) and Excel (Microsoft) software environments.

Unit discharge rate was computed using a time-symmetric algorithm in which the discharge rate was computed from the occurrence of spikes immediately before, after, and during the sampling. We considered a neuron to be sensitive to rotation or translation if the discharge rate was altered by ≥10%. To determine neuronal sensitivity to the angular velocity or the linear acceleration, the discharge rate of a neuron was averaged over ≥12 rotational or translational sinusoids. Averages of motion stimulus and of the discharge rate were fitted with sinusoidal function and the amplitude and phase shift of the fits were estimated. In neurons that could be silenced during the off-direction of rotation, sinusoidal fitting was restricted to periods where linear relationship was observed between the discharge rate and angular velocity. Response nonlinearities due to inhibitory saturation were eliminated by removing parts of the response that deviated from linearity (Chen-Huang and McCrea 1998). The coefficients of fit were obtained after exclusion of the nonlinear part of the response. The ratio of the response amplitude to the peak angular velocity gave an estimate of the rotational sensitivity or gain_R of a neuron (spikes·s⁻¹·deg⁻¹·s⁻¹). The ratio of the response amplitude to the peak linear acceleration gave an estimate of translational sensitivity the response or gain_T of a neuron (spikes·s⁻¹·g⁻¹). The difference between phases of the stimulus and of the response gave an estimate of the response phase shift relative to the angular velocity or linear acceleration of the animal.

Quantitative estimates of responses of various neuronal groups are presented throughout the text as average ± SD. The statistical difference between neuronal subgroup averages was verified using the two-tailed distribution t-test. Goodness of fit was assessed by the determination coefficient (R²) of correlation between the regression model and experimental data.

RESULTS

General characteristics of neuronal activity

The firing behavior of 1,171 neurons in the ventroposterior thalamus of alert squirrel monkeys was studied during whole body rotation and translation in the horizontal plane. In 21 cells the discharge rate was modulated during motion, but this response was attributable to stimulation of somatosensory receptive fields by relative motion of the monkey with respect to the harness restraint system. Thirty-five cells were sensitive to motion in the illuminated room only and their responses were judged to be the result of visual stimulation. One hundred and seven neurons were sensitive to rotational or translational movements in the dark, in the absence of visual and somatosensory stimulation. We classified these cells as vestibular-sensitive thalamic neurons. Except for a few neurons silenced during saccades, none of these neurons was sensitive to eye movements in the absence of head movements.

The background discharge rate in our sample of vestibular sensitive neurons varied between 5.1 and 84.6 spikes/s and was 25.9 ± 15.8 spikes/s on average. The mean rate of background activity observed in squirrel monkey was in the range of that reported for analogous neurons in the rhesus monkey: 10.1 spikes/s (Büttner et al. 1977), 17 spikes/s (Magnin and Fuchs 1977), and 47.4 spikes/s (Meng et al. 2007).

Anatomical reconstruction of the recordings from the ventroposterior thalamus

To identify the locations of vestibular-sensitive neurons in the ventroposterior thalamus we analyzed recordings from identified individual cells and field potentials and examined postmortem frontal sections of the brains with recognizable traces of microelectrode penetrations.

Reconstructions of recordings from the left thalamus of one of the monkeys are presented in Fig. 1, A1 and A2, showing sagittal and frontal views of the area of recordings, respectively. Vestibular-sensitive neurons are depicted as red symbols: circles represent neurons responsive to angular motion, squares represent neurons responsive to linear motion, and diamonds represent neurons responsive to both angular and linear motion. Dark blue circles represent the location of recorded field potential evoked by a flash of light and the location of cells generated spike bursts in response to this stimulus. These blue symbols are concentrated in an area that we identified as the lateral geniculate body (LG). Green circles indicate positions of recordings where vestibulocochlear nerve stimulation evoked recognizable field potentials, but where no neurons sensitive to vestibular stimuli were detected. It is very likely that this area corresponds to the thalamic auditory nucleus—the medial geniculate body (MG). Light blue circles represent cells responding to spatial motion of the animal in the illuminated room, but that were insensitive to the animal's motion in the dark. We suggest that area to be a part of the lateral pulvinar (PuL). Gray circles represent cells with somatosensory input that responded with an instantaneous high-amplitude increase in the firing rate to touch of the hair or gentle pressure onto muscles. In our judgment, the gray symbol corresponds to the somesthetic relay nuclei, lateral and medial ventroposterior (VPL and VPM). Detailed reconstructions of recordings from neurons sensitive to somatosensory inputs are shown in Fig. 1, B–D. Figure 1, B1 and B2 shows sagittal and frontal views of recordings from forelimb-related neurons that were distributed over a large portion of the VP. Figure 1, C1 and C2 shows sagittal and frontal view of recordings from hindlimb- and tail-responsive neurons. These neurons were located primarily posterior and inferior to forelimb-responsive cells. Figure 1, D1 and D2 shows sagittal and frontal view of recordings from head-related neurons that were found in the medial part of the VP. Receptive fields of neurons with various somatosensory inputs are depicted with different symbols, indicated below the panels of Fig. 1.

A photograph of the frontal section of the thalamus of another monkey is shown in the Fig. 1E. Within this part of the section, four microelectrode penetrations had been made. Reconstructed locations of neurons, identified during these penetrations, are indicated with different symbols superposed on the photograph. Red circles represent vestibular-sensitive neurons; white circles represent cells sensitive to animal rotation or translation only in light; and yellow symbols represent neurons with inputs from different parts of the body: diamonds, leg; squares, thigh; circles, tail.

Neurons sensitive to angular and/or linear motion of the animal in the horizontal plane were distributed primarily on anterior and posterior borders of a region that contained neurons responsive to somatosensory stimuli and corresponded to the complex of the ventroposterior nuclei (VP). Within this area the separate clusters of vestibular-sensitive neurons could be distinguished. One cluster was formed by neurons located in the posterior part of the VP, superior to the MG and inferior to the PuL. These neurons were surrounded by vestibular-insensitive cells activated by somatosensory stimuli, such as touch or pressure on the hind limb or tail. Among the vestibular-sensitive neurons studied we found units activated with a short latency of about 4 ms by electrical stimulation of the ipsilateral VIIIn (Fig. 1F). Superior and posterior to this cluster was another group of vestibular-sensitive neurons enclosed by other cells responsive only to visual stimuli. Another cluster of vestibular-sensitive neurons was located in an area corresponding to the oral or superior part of the VP. Among these neurons were cells responsive also to somatosensory stimulation, such as touch of the shoulder or trunk.

Responses to rotation

We recorded from 84 neurons that responded to rotation, which constitute 76% of the entire sample of vestibular-sensitive neurons. They were divided into two groups based on their directional sensitivity: 37 neurons increased their discharge rate during ipsilateral rotation and thus showed type I responses according to the Duensing and Schaefer (1958) classification; 47 neurons showed type II responses by increasing their discharge rate during contralateral rotation. No rotation-sensitive neurons with bidirectional responses were seen. Sixty-six neurons, 62% of the total sample or 78% of rotation-sensitive neurons, responded to rotation only. Eighteen neurons, 17% of the total or 22% of rotation-sensitive neurons, were sensitive to both rotation and translation. A detailed classification of all neurons according to their rotational and translational responses is given Table 1.

Rotational responses were quite variable, both in the amplitude and phase, relative to the animal's motion velocity. The gain_R and phase of the responses of all rotation-sensitive neurons to sinusoids of 0.5-Hz frequency and 40°/s peak velocity are plotted in a polar graph in Fig. 2. To distinguish between the different types of neurons the responsive cells are depicted with various symbols. Type I neurons are shown as diamonds and type II neurons as circles. Dark gray symbols represent neurons responding to rotation only, whereas light gray symbols represent neurons responding to both rotation and translation. To evaluate rotational sensitivity of the population as a whole a mean population response vector in polar coordinates was calculated (Batschelet 1981). The black diamond and circle represent the population mean response gain and phase of type I and type II neurons, respectively. The sensitivity to rotational stimulation did not vary considerably between different groups of neurons. The average sensitivity of all type II neurons (0.25 ± 0.16 spike·s⁻¹·deg⁻¹·s⁻¹) as a group seemed to be greater than that of type I neurons (0.20 ± 0.13 spike·s⁻¹·deg⁻¹·s⁻¹), but the difference was not statistically significant (P = 0.11). Population response sensitivity of both neuronal groups was about 10% smaller than the arithmetic average: it was 0.23 ± 0.20 spike·s⁻¹·deg⁻¹·s⁻¹in type II neurons and 0.17 ± 0.17 spike·s⁻¹·deg⁻¹·s⁻¹ in type I neurons (Fig. 2). Detailed data on responses of various neuronal subgroups are presented in Table 2. In only one case was there a statistically significant difference in sensitivity between subgroups, that is, between type II neurons sensitive only to rotation and type I rotation-sensitive neurons, which were also sensitive to translation (P = 0.04). Because of the similarity in the response magnitude across neurons of various types we pooled together all of the data and calculated the average sensitivity of the entire population of rotation-sensitive cells: 0.23 ± 0.15 spike·s⁻¹·deg⁻¹·s⁻¹.

For rotation-sensitive neurons as a group the response phase relative to on-directional angular velocity was −9.3 ± 34.1°. The response phase varied widely, ranging from a phase lead of 73.7° to a phase lag of −88.3°. This variability is reflected in the high SD of the phase mean. Average responses of type II neurons had a greater phase lag (−16.4 ± 32.03°) than that of responses of type I neurons (−1.81 ± 36.82°), but the difference fell just short of statistical significance (P = 0.06). On average, the responses of type I and type II neurons sensitive to both rotation and translation lagged behind the angular velocity more than neurons sensitive to rotation only (Table 2). However, due to the high variability in phase shifts, the differences between all subgroups were statistically insignificant (P > 0.05).

Rotational responses in dark and light

We considered neurons to have vestibular input if they responded to angular or linear motion in darkness. To clarify whether the vestibular responses were modified by illumination, we compared rotational responses in light to responses in the dark in 32 of 84 rotation-sensitive neurons.

The phases of the rotational responses of these neurons in the light are plotted against the phases of responses in the dark (Fig. 3C). The thick diagonal line indicates zero difference in phase, whereas two thin diagonal lines show phase shifts of +45° and −45°. The gain of rotational responses in the light versus the gain of responses in the dark is plotted in Fig. 3D. The diagonal line in this panel indicates constancy in the gain for responses under the two conditions.

Two groups of neurons could be distinguished on the basis of the phase shifts of their responses in the dark versus the illuminated room.

In one group we included neurons (27 of 32, 84%) whose responses to rotations in the light and in the dark either did not change their phases or the phase lead of responses in the light increased relative to the phase lead of rotational responses in the dark. Examples of representative neuron responses to rotations in the dark and in the light are shown in Fig. 3, A1 and A2, respectively. In 18 neurons response phases in light deviated from response phases in the dark in the range of −6.9–21.2°; however, statistically those differences were indistinguishable (P > 0.05). In 9 neurons the responses in the light increased their phase lead by 26.2–56.3 ° and these shifts were statistically significant (P < 0.01). Despite the variability, all these neurons represented a continuum with respect to changes in phase of rotational responses in the darkness and in the illuminated room (Fig. 3C, gray symbols). For all these neurons pooled together, the phase lead of responses in the light exceeded, on average, by 14.2 ± 21.4° the phase of responses in the dark, although this difference was statistically insignificant (P = 0.13). Rotational sensitivity was not altered with illumination: the gain for responses in the light was not different from the gain for responses in the dark, deviating by only 0.03 ± 0.12 spike·s⁻¹·deg⁻¹·s⁻¹, on average, and the difference was not significant (P = 0.61) (Fig. 3D, gray symbols).

Another, smaller group consisted of neurons (5 of 32, 16%) that responded to rotations in the light in a different way. The responses of one of these neurons to rotations in the dark and in the light are shown in Fig. 3, B1 and B2, respectively. These neurons were distinguished by a substantial increase in the phase lag of their responses in the light compared with their responses in the dark. In each of these neurons the difference between phases of the responses was statistically significant (P < 0.01 in four neurons; P < 0.025 in one neuron). On average, the phase shift of responses in the dark and in the light was −52.6 ± 12.3° (Fig. 3C, black symbols). Phases of averaged responses to rotation in two distinct conditions differed significantly (P < 0.01, paired t-test with two-tail distribution). Rotational sensitivity in the light was not higher than that in the dark, although their average estimates diverged by 0.11 ± 0.15 spike·s⁻¹·deg⁻¹·s⁻¹; the difference was not statistically significant (P = 0.22) (Fig. 3D, black symbols).

The constancy in rotational sensitivity in the major group of neurons is an indication that in the thalamus, the vestibular signal is processed independently from visual signals. The small increase in responsiveness observed in a part of these neurons during rotation in light more likely reflects an increase in background activity due to a general elevation of sensory inputs in the illuminated environment, rather than a response to specific visual stimuli.

A significant increase in the phase lag of responses in the light, seen in a small group of rotation-sensitive neurons, may reflect a transition in which their responses instead of being related to rotational velocity began to relate to position. These neurons were nested among visually sensitive cells and were primarily located superior and posterior in the explored area, which we identified as the PuL. It is very likely that they represent pulvinar visually sensitive cells that also receive vestibular input. One result of the vestibular input to these cells may be an increase in their sensitivity to the visually perceived movement of surroundings during self-motion.

Rotational response as a function of velocity

In 15 neurons we recorded responses to rotations for varying velocities, which ranged from 4 to 100°/s. In all these neurons we saw that the response amplitude increased with the rotational velocity. This increase, however, was not linear. A ratio of the response amplitude to the amplitude of the rotational stimulus, defined as response gain_R, decreased with increasing stimulus velocity. Figure 4 illustrates the changes in the absolute and relative magnitudes of the responses to 1-Hz rotational sinusoids. Figure 4, A–E shows responses of a neuron that were symmetrical during excitatory and inhibitory parts of the cycle. Figure 4, F–H shows responses of a neuron with inhibitory saturation during rotation with the highest of used peak velocities. The rotational gain_R of the 15 neurons is plotted as a function of the peak angular velocity in Fig. 4I. The data are modeled by the power function shown as a gray line. The fit yields an estimate of average gain_R = 7.8 × (velocity)^−0.9, R² = 0.54. To simplify the equation, we set an exponent (power) to negative unity and obtained gain_R = 11.3 × (velocity)⁻¹, which can be written as gain_R = 11.3/velocity. It follows from this ratio that for the range of velocities tested, the correlation between gain_R and angular velocity appears to be quite straightforward: rotational sensitivity is inversely proportional to the velocity of rotation.

The phase of responses to rotations with different angular velocities did not vary consistently in the cells we studied (Fig. 4J).

Rotational response as a function of frequency

In nine neurons the rotational responses were tested with two to five frequencies in the range of 0.2–4.0 Hz. The responses of one of those neurons are shown in Fig. 5, A–C. The response amplitude increased with the rotational frequency. Rotational sensitivity, or gain_R, also increased with the rotational frequency. This increase in gain_R was not linear. Gain_R is plotted as a function of the rotational frequency in Fig. 5.

The phase of rotational responses of different neurons varied substantially between a phase lead of 90° to a phase lag of −60° relative to the velocity of rotation. In most cases, the responses slightly led the rotational velocity or were in phase with it. There was no significant change in the phase of responses as the rotational frequency increased (Fig. 5E).

Responses to translation

Forty-one neurons (38% of our entire sample) were sensitive to linear translation. Twenty-three of these neurons (56%) responded to translation only and 18 of them (44%) responded to both translation and rotation. We did not study the spatial distribution of translational responses, and we restricted testing to translations along two cardinal axes: interaural and nasooccipital. Sixteen cells (39%) responded to translations along the interaural axis. Twenty-two neurons (54%) responded to translations along the nasooccipital axis. The latter group, which included six neurons responsive to the second harmonic of translational sinusoid, were of type III according to the classification of Duensing and Schaefer (1958). All other neurons showed unidirectional response preference. Three neurons (7%) responded similarly to translation along both axes. A detailed classification of these translation-sensitive neurons is given in the Table 1.

Sensitivity (gain_T) and phase shift of translational responses were calculated relative to the peak acceleration of sinusoidal motion. Figure 6 uses polar coordinates to display the gain_T (Fig. 6A) and phase (Fig. 6B) of responses to interaural and nasooccipital translations with 0.5-Hz frequency and 0.1 g peak acceleration. The gain_T and phase of the population responses of subgroups of neurons sensitive to ipsi- or contralateral interaural translation, and fore or aft nasooccipital translation, were calculated as polar coordinates of a mean vector (Batschelet 1981). The arithmetic average gain_T and phase were calculated to estimate the variability across the responses of individual neurons. The results of this analysis of the different subtypes of the translation-sensitive neurons are summarized in Table 3.

The translational response gains were similar in neurons sensitive only to translation and in those sensitive to both translation and rotation. In addition, they were similar among neurons showing different directional sensitivities to translation. The only differences in average gain_T were between neurons responsive only to translation along the interaural axis and neurons responsive to both translations along the interaural axis and to yaw rotation (Table 3). These differences, however, were statistically insignificant (P = 0.17) and probably result from the small sample size. Because of this similarity in translational sensitivity, we grouped the data together and calculated the average sensitivity of the whole sample of translation-sensitive neurons to be 111.3 ± 53.7 spikes·s⁻¹·g⁻¹.

The phase of the responses to translation varied significantly among neurons, but in general, translational responses lagged behind peak stimulus acceleration. Estimates of response phase shifts, which are shown in Table 3, were calculated relative to on-directional acceleration of linear motion. No statistically significant differences were found in comparing the averaged response phases to the population response phases of neuronal subgroups sensitive to translation only or to both translation and rotation. The same was found in a comparison of the averaged response phases to the population response phases of neuronal subgroups sensitive to translation along one of the cardinal axes. On average, the response phase shift in all translation-sensitive neurons relative to on-directional translation was −71.9 ± 42.6°. This result shows that translational responses of thalamic neurons lagged behind the acceleration of linear motion, but led the linear motion velocity.

In nine neurons we tested responses to translations with various frequencies and accelerations. Exemplary responses from one of these neurons are shown in Fig. 7, A–C. The gain_T in this neuron decreased with increasing acceleration. The same was true in the other cells and is visible in the plot of gain_T versus translational acceleration (Fig. 7D). The relationship between gain_T and translational acceleration was not linear and was best approximated with a power function. The data are fit with the equation gain_T = 53.7 × (acceleration)^−0.49, R² = 0.26 (Fig. 7D, solid gray line). In the equation the exponent (power) is equal to −0.5, which means that the sensitivity to translational acceleration was inversely proportional to the square root of the acceleration. Bearing in mind that acceleration is proportional to the square of velocity we can assume that translational gain is inversely proportional to linear stimulus velocity and thus is similar to what we have found in the relationship between the neuronal rotational gain and stimulus angular velocity. In contrast, the phase of the translational responses did not vary with the acceleration in the neurons tested (Fig. 7E).

Convergence of rotational and translational inputs

Among neurons sensitive to vestibular stimulation were 18 cells (17% of our entire sample) that responded to both rotational and translational stimulation. These cells showed no directional preferences in their responses. They were almost equally divided into neurons with type I and type II rotational responses. The fraction of these neurons sensitive to nasooccipital translations was equal to the fraction of neurons sensitive to interaural translations. Details of these results are listed in Table 1.

Responsiveness to angular and linear motion varied between these neurons. We did not find that the sensitivities to rotational and translational stimuli varied in direct proportion with each other.

DISCUSSION

In our study of ventroposterior thalamus neurons sensitive to vestibular stimulation we concentrated on cells that were responsive to rotation and/or translation in the horizontal plane only. We found these neurons to be distributed across a wide area in the ventroposterior thalamus. Our observations are in agreement with those of Büttner and colleagues (1977), Magnin and Fuchs (1977), and Meng and colleagues (2007), who reported that the neurons sensitive to rotation in rhesus monkeys were scattered within the limits of the ventroposterior nuclei and were intermingled with other cells that did not receive vestibular input. Our data are also in accord with the results of Liedgren and colleagues (1976), who studied neuronal activity in the ventroposterior thalamus in the squirrel monkey. They found that neurons responsive to vestibular nerve stimulation were not confined to one cell group, but were dispersed among the various somatosensory nuclei, mostly in the oral part of pulvinar, and the oral and caudal parts of the VPL, whereas no vestibular-sensitive neurons were detected in the VPM or in the central part of the VPL.

We observed that the vestibular-sensitive neurons in the ventroposterior thalamus were not randomly distributed; instead, vestibular-sensitive neurons were grouped into separate clusters. One cluster was located within the midst of adjoining areas that we recognized as the posterior region of the VP, an area of the PuL, and the superior part of the MG. This cluster therefore coincides with the inferior part of the nucleus ventralis posterior, pars posterior (VPP), identified by Akbarian et al. (1992) as the main source of thalamic projections to the parietoinsular vestibular cortex (PIVC) in the squirrel monkey. This cluster is very likely to be analogous to the “vestibular cluster” of large, multiangular cells emerging from the dorsal aspect of MG in the macaque monkey (Craig 2004). Another distinct cluster of vestibular-sensitive neurons was situated superior and posterior to the one described earlier. We suggest that these neurons, which were surrounded by cells sensitive only to visual stimuli, were located in the PuL. A third distinctive cluster of vestibular-sensitive neurons was observed in the area that we classified as the superior border of the VP. This observation is in an agreement with the data of Liedgren et al. (1976), who identified in the oral part of the VPL a majority of cells combining short-latency vestibular responses and somatosensory responses from proximal parts of the limbs and trunk. It is possible that these different clusters of vestibular-sensitive neurons in the ventroposterior thalamus correspond to distinct, separate sources of projections from the thalamus to different cortical areas involved in the processing of vestibular signals, as described by Akbarian et al. (1992). Using retrograde labeling of thalamocortical neurons with horseradish peroxidase in squirrel monkeys, these researchers found that the PIVC received input mainly from the posterior part of ventroposterior nuclei (VPP), the somatosensory cortical area 3aV received input mainly from the superior region of ventroposterior nuclei, and the association parietotemporal cortical area T3 received input mainly from the oral and medial pulvinar.

Most of our vestibular-sensitive neurons responded to rotation in the horizontal plane. The average rotational sensitivity of these neurons, which was estimated for responses to 0.5-Hz sinusoids with 40°/s peak velocity, was 0.23 spike·s⁻¹·deg⁻¹·s⁻¹.

We found that despite the extensive variation between neurons the mean response phase generally followed changes in the angular velocity. This echoes studies of the phase shift of thalamic neuron responses in the rhesus monkey (Büttner et al. 1977; Magnin and Fuchs 1977; Meng et al. 2007). There are, however, some quantitative differences between those studies and ours. In the rhesus monkey experiments the response had been found to lead the angular velocity by 10–30°, whereas we found a small phase lag of −9.3° for the response in the squirrel monkey. It is reasonable to compare response phase shifts of thalamic neurons with those recorded from vestibular nuclei neurons. In the gerbil brain stem it was found that vestibular-sensitive neurons responded to 0.1- to 2-Hz sinusoids with 60°/s amplitude in phase with rotational velocity (Kaufman et al. 2000). In alert rhesus monkeys, the rotational responses of vestibular nuclei neurons lead the angular velocity; three different studies in monkeys have produced very similar estimates of response phase lead: 2–35° over the frequency range 0.2–0.93 Hz (Fuchs and Kimm 1975), ≤10° at frequencies >0.2 Hz (Buettner et al. 1978), and 20 ° at 0.5 Hz (Dickman and Angelaki 2004). According to the data from the rhesus monkey, the mean thalamic neuron response to rotation shows a phase lead that exceeds the response of vestibular nuclei neurons. This difference in response phases suggests that the rotational signal is rapidly transmitted between the vestibular nuclei and the thalamus. In contrast, our data indicate that the thalamic neuron response has a small lag behind the vestibular nuclei neuron response, suggesting that a small delay occurs in processing signals in the vestibulothalamic pathway.

In the available literature examining rotational responses of thalamic neurons we did not find reports on the relationship between rotational sensitivity and angular velocity. Büttner et al. (1977) analyzed gain as a function of the stimulus frequency. In their experiments, however, rotations with different frequencies were performed with different angular velocities. If in the plots of their study the rotational velocities are substituted for the corresponding frequencies it can be seen that the response gain decreased with increasing peak velocity. A similar relationship between rotational sensitivity and the angular velocity is seen in the vestibular nuclei neurons. For example, Schneider and Anderson (1976) found that the response gain in gerbils decreased with increasing angular velocity and, in the most sensitive neurons, the response gain showed an almost logarithmic decrease with increasing stimulus velocity. Dickman and Angelaki (2004), who analyzed response gain as a function of stimulus frequency, found that the increase in gain followed an increase in frequency. In fact, at frequencies where response gain increased, the peak angular velocity was not the same, but decreased from 30 to 15 and 7.5°/s, thus indicating the decrease in gain with the increase in rotational velocity. We observed that the phase of responses to rotations with different angular velocities but fixed stimulus frequency did not vary consistently. A similar constancy in response phase during rotational stimulation at different velocities was observed in thalamic units of rhesus monkeys (Magnin and Fuchs 1977). These results suggest that the temporal relationship between stimulus and rotational response of thalamic neurons is not affected by the speed of rotational movement, at least within the range of angular velocities examined.

Our results on the relationship between response to rotation and stimulus frequency are in a good agreement with the data of Büttner et al. (1977) and Meng et al. (2007), who studied the rotational responses of thalamic neurons in rhesus monkeys. Their estimate of the absolute gain was 0.56 spike·s⁻¹·deg⁻¹·s⁻¹ at 0.1 Hz and 0.54 spike·s⁻¹·deg⁻¹·s⁻¹ at 0.5 Hz. They reported that for their sample of neurons, the responses were in phase with or led the stimulus velocity by 10–30°. The relationship between the response gain and rotational frequency in the range of 0.2–4 Hz for the thalamic neurons in the squirrel monkey (this study) and those recorded in the rhesus monkey (Büttner et al. 1977; Meng et al. 2007) are generally similar to what has been seen in the vestibular nuclei neurons in gerbils (Schneider and Anderson 1976) and in monkeys (Buettner et al. 1978; Dickman and Angelaki 2004). In both vestibular nuclei and thalamic neurons, the response gain increased with the stimulus frequency. A substantial difference between these two groups of neurons is the magnitude of their rotational sensitivity. In our sample of thalamic neurons, the average gain of 0.45 spike·s⁻¹·deg⁻¹·s⁻¹ at 1 Hz was about two times lower than that of the vestibular nuclei neurons recorded by Buettner et al. (1978), who found it to be 0.77 spike·s⁻¹·deg⁻¹·s⁻¹, or Dickman and Angelaki (2004), who found it to be 0.9 spike·s⁻¹·deg⁻¹·s⁻¹ at 1-Hz frequency. In both thalamic and vestibular nuclei neurons the rotational responses tended to increase their phase lead relative to angular velocity, but this tendency did not reach statistical significance for either type of neuron.

Our estimate of the mean translational sensitivity of thalamic neurons (111 spikes·s⁻¹·g⁻¹) was almost identical to the average sensitivity (103.8 spikes·s⁻¹·g⁻¹) estimated for thalamic neurons sensitive to translation in the rhesus monkey (Meng et al. 2007). A difference in translational sensitivity had been found between vestibular nuclei neurons with only otolith input and those with both otolith and canal inputs: the gain of the translational responses of the latter type of neurons (219 spikes·s⁻¹·g⁻¹) was almost twice as high as that calculated for the former (134 spikes·s⁻¹·g⁻¹) (Dickman and Angelaki 2002). In our sample of thalamic cells, we did not see significant differences in sensitivity between neurons responsive to translation only and neurons responsive to translation and rotation. In acute and chronic experiments in cats and monkeys it has been found that responses of vestibular nuclei neurons to activation of otolith afferents were in phase with the velocity of motion (Dickman and Angelaki 2002; Schor et al. 1998; Tomlinson et al. 1996). In a detailed study Dickman and Angelaki (2002) reported that translation-sensitive vestibular nuclei neurons could be divided into three groups according to their response dynamics: the most common group showed a responses phase lag of less than −60° relative to acceleration, another group showed a lag between 0 and −50°, and a third group showed a modulated firing rate almost in phase with acceleration. Taken together, these findings indicate that the phases of translational responses are distributed between stimulus acceleration and velocity and, on average, correlate more tightly with velocity. A similar temporal relationship between neuronal response and translational stimulus is seen during processing of motion signal in vestibular-sensitive thalamic neurons.

A convergence of rotational and translational inputs was seen in 17% of the vestibular-sensitive thalamic neurons we studied. We suggest that a small proportion of neurons with convergence of signals from different end organs, which were identified in our sample of cells, was the result of limitation of motion to the horizontal plane only. For comparison, Meng and colleagues (2007) reported that 90% of thalamic neurons they recoded in the rhesus monkey were sensitive to both rotational and translational stimuli. In the vestibular nuclei the number of neurons sensitive to both rotational and translational stimuli has differed from study to study. In decerebrated cats a convergence of both inputs was found in 17–21% of all cells recorded, according to Zhang et al. (2001) and Schor et al. (1998). In the alert gerbil a convergence of the inputs was seen in 38% of neurons (Kaufman et al. 2000). Fully half of the neurons recorded in the alert rhesus monkey vestibular nuclei have been reported to be sensitive to both motion stimuli (Dickman and Angelaki 2002; Tomlinson et al. 1996).

In our experiments, we found that sensitivities to translational and rotational stimuli were independent of each other; moreover, high sensitivity to one input did not necessarily mean high sensitivity to the other. This could result from spatial separation of the canal and otolith organ inputs from the vestibular nuclei. The former have been found primarily in the superior and lateral vestibular nuclei, whereas the latter have been found more posterior in the lateral, medial, and descending nuclei; both inputs overlap in the lateral vestibular nucleus (Zhang et al. 2001). In the thalamic neurons that receive projections from the midst of that overlap, the inputs from canal and otolith organs could be equal, whereas in other thalamic neurons one of the inputs could be larger than the other.

Our results imply that changes in the discharge rate of thalamic neurons, which code motion in the horizontal plane, can be approximated with power functions of the velocity of displacement. Such a compressed functional relationship between neuronal response and stimulus is common in the sensory systems (Drew and Abbott 2006; Johnson et al. 2002; Stevens 1970). These power relationships have been considered to be an effective strategy for neuronal responses to encode stimuli that vary across a wide dynamic range (Dayan and Abbott 2001; Stevens 1970). Since the range of self-motion velocities overlaps four decades in the decimal scale, the demands are high for vestibular-sensitive neurons to effectively encode signals of motion. The idea that neuronal responses to sensory stimuli can be expressed as a power law correlates with quantitative measures of how these stimuli are cognitively perceived. The psychophysical law of Weber, as modified by Stevens (1961), postulates that sensation is a power function of the stimulus magnitude. The overall efficiency of the relationship between neuronal activity coding spatial motion and the subjective perception of that motion from the vestibular activity is quite effective: the accuracy in subjective estimates of translations and rotations in the horizontal plane have been found to closely match the actual magnitude and velocity of the spatial displacement (Becker et al. 2000; Israel et al. 1997; Jurgens et al. 1999; Marlinsky 1999a,b; Mittelstaedt and Mittelstaedt 2001).

GRANTS

This research was supported by National Institute on Deafness and Other Communication Disorders Grant DC-05056.

FOOTNOTES

• The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.