| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Division of Biology, California Institute of Technology, Pasadena, California 91125; and 2Departments of Physiology & Pharmacology and Psychology, University of Western Ontario, London, Ontario N6A 5K8, Canada
Submitted 9 January 2004; accepted in final form 28 February 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Lackner and Dizio 2000; Mergner et al. 1997). An additional predominant source of kinesthetic input is the muscle spindle organ, which relays both muscle length and velocity (Matthews 1972). One way to demonstrate the contribution of muscle spindles to kinesthesia is to mechanically vibrate muscle bellies or tendons. Provided the vibration frequency is within a physiologically relevant range, the entrainment of muscle spindle discharge to the vibration frequency results in a false signal of muscle lengthening (Matthews 1982; Roll and Vedel 1982). Numerous experiments over the past three decades have demonstrated that such a false signal can induce profound distortions in the body schema (Craske 1977; Goodwin et al. 1972; Lackner 1988), suggesting the proprioceptive information provided by muscle spindles plays a central role in the kinesthetic construct.
The distribution of muscle spindles throughout the body is remarkably heterogeneous (Voss 1971). From a teleological perspective, muscles with higher spindle content presumably subserve important sensory roles requiring a higher resolution. For example, the spindle content in distal finger muscles, among the highest in the body, likely relates to their role in active, exploratory touch. Surprisingly, a similarly high density of muscle spindles is observed in the deep muscles of the neck (Cooper and Daniel 1963; Richmond and Abrahams 1975; Richmond and Bakker 1982), and marked heterogeneities in spindle distribution are even observed within the same neck muscle (Richmond et al. 1999b). The microstructure and organization of neck muscle spindles appear to provide a highly resolved sense of head-on-body position. Accordingly, disruption of neck muscle afferent information induces profound disruption to not only balance and posture, but also to guided actions (Biemond and De Jong 1969; Cohen 1961; De Jong et al. 1977). Indeed, because many major exteroreceptive sense organs (e.g., eyes and ears) are mounted within a mobile head, knowledge of the position of the head relative to the body is required for many sensorimotor transformations.
Neck muscle afferent information is susceptible to alterations through mechanical vibration. Numerous human studies have documented influences of neck muscle vibration on balance, orientation or posture (Bove et al. 2002; Ivanenko et al. 1999; Karnath et al. 1994, 2002; Kavounoudias et al. 1999; Lackner and Levine 1979; Lewald et al. 1999; Lund 1980; Popov et al. 1996) as well as on pointing movements (Biguer et al. 1988; Roll et al. 1991; Taylor and McCloskey 1991). The typical result is that unilateral vibration produces horizontal deviations in action or perception, consistent with the illusion of a head turn in the direction that would lengthen the muscle(s) being vibrated. Neck muscle vibration also leads to the illusory percept of motion of a light in an otherwise dark room (Biguer et al. 1988; Roll et al. 1991; Taylor and McCloskey 1991), secondary to slow-phase eye movements in the opposite direction, likely as a result of the cervico-ocular reflex (COR; McKenna et al. 2003; Popov et al. 1999). The myriad effects of neck muscle vibration on action argue that proprioceptive signals are fundamental for the sense of head-on-body position for perception and action.
The goal of this paper was to develop an animal model of some of the behavioral influences of neck muscle vibration. Development of such a model is a necessary prerequisite for a mechanistic understanding of how afferent information from neck muscles, and perhaps any muscle, is integrated into neural plans for action. Such a model may also provide insights into the neural mechanisms by which neck muscle vibration exerts therapeutic value in a variety of pathological conditions (Karnath et al. 1993, 2000; Schindler et al. 2002). We have elected to examine the effects of neck muscle vibration on the saccadic system of rhesus monkeys for a number of reasons. Over 40 yr of research in rhesus monkeys have yielded an excellent understanding of the sensorimotor transformations underlying saccadic eye movements (Chalupa and Werner 2004), and the posterior parietal cortex in particular appears to have a primary role in integrating head position sensation for saccades (Brotchie et al. 1995, 2003; Snyder et al. 1998). Rhesus monkeys can also be trained to perform complex behavioral tasks necessary to assess the effects of neck muscle vibration. Here, we employ the endpoints of memory-guided saccades as a proxy for the influences of neck muscle vibration on the oculomotor system. Such saccades require the monkey to hold the location of a flashed visual cue in memory, and only look to the remembered location of this target after an imposed delay interval has expired. Our approach is to apply neck muscle vibration unilaterally to the dorsal neck during the delay interval on one-half of the trials (vibration saccades) and compare the characteristics of such vibration saccades to control saccades in which vibration was not applied during the delay interval.
Some results have been reported previously in abstract form (Corneil and Andersen 2003a,b).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Two adult male rhesus monkeys (Macaca mulatta), weighing 7.5 and 10.5 kg, were used in these experiments following protocols approved by the Caltech Institutional Animal Care and Use Committee and in compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Both monkeys underwent an aseptic surgery during which a scleral coil was implanted subconjunctivally (Judge et al. 1980) for monitoring the horizontal and vertical rotation of the eyes in space (henceforth referred to as eye position; Fuchs and Robinson 1966; torsional eye rotations were not measured), and a head post was secured to the skull by way of a dental acrylic pedestal for immobilizing the head during the experimental sessions. Anesthesia was induced with ketamine hydrochloride and maintained with isoflurane. Antibiotics were administered pre- and postoperatively, and anti-inflammatories and analgesics were administered postoperatively. The monkeys' weights were monitored daily before and after surgery, and their general health was supervised by the university veterinarian.
Prior to the start of an experimental session, the monkeys were placed within a primate chair, and their heads restrained via the implanted head post. Within the chair, the monkeys rested on a perch with their trunk and hips facing forward. The monkeys were placed within a dark, sound-attenuated experimental chamber, in the center of magnetic fields produced by 1-m field coils (CNC Engineering). The monkeys faced a hemispheric array of red light emitting diodes (LEDs; each 4.7 cd/m2) subtending about ±40° of the central visual field. These LEDs were arranged radially around a central LED, the fixation point (FP), which was located 50 cm straight ahead of the monkey. Horizontal and vertical eye positions were each recorded at 1,000 Hz via a PLEXON system (Plexon, Dallas, TX), and all aspects of the experimental paradigms were controlled by customized real-time Labview programs interfacing with the hardware through a PXI box (National Instruments). The monkeys were monitored throughout the experiment via an infrared camera and light-source, positioned behind the monkeys so that the infrared light source was not in their field of view.
Experimental paradigms
Monkeys were trained to perform a delayed-memory saccade task for a liquid reward. This task requires the monkey to look to the remembered location of a flashed target only when instructed (Fig. 1 A). Trial onset was signaled by the removal of a background diffuse white light (
1.0 cd/m2), which was flashed between trials to prevent dark adaptation. The central FP was illuminated 250 ms later, and the monkeys were required to look at the FP within 1,500 ms. The monkeys maintained fixation on the FP within a circular window of 3° radius for 500–1,500 ms and kept looking at the FP while a peripheral target was briefly flashed for 350 ms. To receive the liquid reward, the monkeys had to look to the remembered location of the flashed cue within 1,000 ms only after the central FP was extinguished, which constituted the "go" signal (Fig. 1A). The interval between the offset of the cue and the go signal defined the delay interval. Once entering the window around the cue, the monkeys had to maintain fixation for a subsequent 500 ms before receiving a liquid reward. To prevent the monkeys from learning whether their memory-guided saccades were inaccurate, the target was not re-illuminated after the monkey looked into the target window.
|
; Gnadt et al. 1991; White et al. 1994). To accommodate for this, the target window could be as large as 10° and shifted upward by 
10° relative to the veridical target location.
The goal of this experiment was to compare the endpoints and metrics of memory-guided saccades generated with or without neck muscle vibration. To permit a sufficient amount of time for the effects of neck muscle vibration to accumulate, the monkeys were initially trained to generate memory-guided saccades following delay intervals that spanned 4,000 ms. Once proficient at this task, the monkeys were acclimatized to neck muscle vibration delivered to the muscles of the dorsal neck. Vibration was generated by a commercially available permanent magnet shaker (model LDS-203, Ling Electronics) secured to a horizontal neck plate above the monkeys' shoulders. A contact surface consisting of a customized plexiglass hemisphere of 1.5 cm diam was placed in contact with the monkeys' dorsal neck, generally between 2 and 5 cm below the external occipital proturberance and 2 and 4 cm lateral of the dorsal midline. Contact was made on either the left or right side of the dorsal neck on alternating days. The shaker was driven by a custom-designed feedback amplifier capable of driving vibration peak-to-peak amplitudes over a range of 50–1,000 µm at frequencies ranging between 30 and 250 Hz. However, neck muscle vibration never exceeded 300 µm or 125 Hz. Vibration duration was controlled by the experimental software and ranged between 500 and 3,500 ms. The two monkeys acclimatized to vibration quite rapidly and began making memory-guided saccades in the presence of vibration on the first day it was introduced.
The animals performed memory-guided saccades in which vibration was applied on one-half of all trials (vibration trials and vibration saccades) and not applied on the other half (control trials and control saccades). The standard parameters of vibration were 1500 ms in duration, 200 µm in amplitude, and a frequency of 100 Hz. On vibration trials, vibration began 500 ms into the delay interval and persisted until the reward was delivered (Fig. 1A) and was therefore ongoing when the memory-guided saccade was generated. The duration of the delay interval was set to be equal in both control and vibration trials (i.e., the delay interval for control trials was 500 ms longer than the vibration duration). Within a block of trials, targets could be located at either two or four potential locations that were always placed at an eccentricity of 10°. When two potential cues were used, they were either placed at 10° left and right or 10° up and down. When four potential cue locations were used, they could either be located at the four cardinal locations or at four oblique locations located at polar angles of 45, 135, 225, or 315° (with 0° designating straight right). An experimental block consisted of about 30 successful trials of each unique combination of trial type (vibration vs. control) and target location. All trial combinations were presented an equal number of times within an experimental block by completing subblocks of 10 correct trials of each trial combination. Within each subblock, the upcoming trial was randomly selected from the remaining trial combinations required to complete the subblock.
Portions of the dataset consisted of three sequential experimental blocks wherein one of the vibration duration, amplitude, or frequency was changed to one of three values, while the other two vibration parameters stayed at the standard levels. It was necessary to use a blocked design because we could not adjust the amplitude or frequency of vibration on-line. To avoid sequence effects, on a given day, the monkey completed one of the six possible permutations of a low-medium-high sequence and over six successive experimental blocks, each permutation was presented an equal number of times.
Data analysis
Our approach was to compare the properties of control and vibration saccades generated within a block of trials. Figure 1B graphically depicts the parameters that will be used to establish the effects of neck muscle vibration, beginning with a comparison of the spatial endpoints (measure 1 in Fig. 1B) of vibration and control saccades. Across vibration and control saccades, we also compared the amplitude of the horizontal and vertical component (measure 2 in Fig. 1B), the horizontal and vertical position of the eye at the onset of the memory-guided saccade (measure 3 in Fig. 1B, taken relative to the average eye position over the interval from 100 ms prior to 50 ms after cue onset), the radial directions (measure 4 in Fig. 1B), the stability of the eye during the delay interval (measure 5 in Fig. 1B), and the saccadic reaction time (SRT) of memory-guided saccades relative to the go signal (measure 6 in Fig. 1B).
To group the datasets across the side of application, the horizontal component of some of these parameters (e.g., spatial endpoints, saccade amplitude) will be flipped if vibration was applied to the right side of neck. Thus all results presented in this paper are presented as if vibration was applied to the left side of the neck.
Off-line, the onset and offset of memory-guided saccades were detected by a custom-made MATLAB program using a velocity criteria of 50°/s. Position traces were first smoothed with a Savitzky-Golay filter (3rd order, 11-point frame size). Instantaneous eye velocities were derived from the position traces using a two point differentiator and smoothed with a low-pass Butterworth filter (fs/fc ratio = 17). All markings were checked by the experimenter and corrected if necessary. SRTs <80 ms were excluded since these saccades were presumably anticipatory, and SRTs >600 ms were excluded because of a presumed lack of alertness. Trials were also rejected if the direction of the first saccade was not within ±90° of the target direction. Overall, <1% of all trials were excluded using these criteria.
We analyzed ocular stability during the delay interval by counting the number of microsaccades and deriving the mean drift velocity. These parameters were measured by first deriving de-saccaded velocity profiles by cutting out any points where the vectorial eye velocity exceeded 50°/s. The velocity crossing threshold for microsaccades was identified as being >3 SD away from the mean velocity of this de-saccaded profile over the entire trial for 
5 ms. These microsaccades were removed, and a least squares regression line was fit to the remaining eye position traces. The mean drift velocity during the delay interval was taken as the slope of this regression line.
To analyze the difference between spatial endpoints, a two-dimensional (2D) Kolmogorov-Smirnov (KS) statistic for the difference between distributions was determined (Press et al. 1992). This statistic is based on the cumulative probabilities constructed from the ranked data arrays broken down into four quadrants and is particularly useful for limited numbers of data points for which the underlying probability distributions are unknown. Other conventional parametric and nonparametric analyses were applied depending on the distribution and origin of the datasets.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
We first examine the influence of neck muscle vibration applied to the left side of the neck on memory-guided saccades, using standard vibration parameters (amplitude = 200 µm, frequency = 100 Hz, duration of vibration = 1,500 ms). Figure 2 displays the results from a representative block of trials, presenting a x-y-plot of the endpoints of control and vibration saccades directed to targets located 10° to the right or left. In this example, vibration saccades end elevated and displaced slightly leftward compared with control saccades. A statistical test (2D KS test) confirmed that the differences in the endpoint distributions for vibration and control saccades were significant in this experimental block for both target directions (P = 0.01 to the right, P = 0.04 to the left). A comparative analysis of the variance of leftward control and vibration saccades in both the horizontal and vertical dimensions revealed little influence of vibration (horizontal SD = 0.80° for control saccades, 0.89° for vibration saccades; vertical SD = 1.68° for control saccades, 1.66° for vibration saccades) and only moderate influences for the right target (horizontal SD = 1.42° for control saccades, 1.14° for vibration saccades; vertical SD = 1.27° for control saccades, 1.37° for vibration saccades).
|
|
Our hypothesis assumes that neck muscle vibration acts through altered neck muscle afferent activity, and accordingly, it is plausible that vibration may induce eye movements through the cervico-ocular reflex (COR). We therefore examined the stability of the eye during the delay interval, comparing across vibration and control trials both the number of small microsaccades and the horizontal and vertical drift velocities of the desaccaded eye position traces.
Figure 4, A and B, shows examples of vertical eye position traces during individual control and vibration trials. Two features of vibration trials are apparent from these traces: an increased upward drift of the eye and an increased incidence of downwardly directed microsaccades. Such features were observed consistently: the mean frequency of microsaccades was significantly greater in vibration trials than control trials (Fig. 4C, paired t-test, P < 10–10), and the upward drift of the eye during the delay interval was significantly greater during vibration than control trials (Fig. 4D, paired t-test, P < 10–6). A similar analysis of the horizontal drift velocity revealed a trend toward greater drift of the eye away from the side of vibration on vibration trials (paired t-test P < 0.02), although the accentuation of this drift was not as great as in the vertical domain.
|
The observation of an augmented upward eye drift during the delay interval of vibration trials raises the concern that the differential distribution of saccadic endpoints could result simply because the monkeys generate saccades with exactly the same metrics that differ only in starting position. To investigate this, we performed three analyses. First, we compared the horizontal and vertical saccadic amplitudes of both control and vibration saccades. Figure 5 plots the differential distribution of saccadic amplitudes for both vibration and control saccades, using data from the same block as in Fig. 2. Similar to the analysis of saccade endpoints, vibration saccades had a greater vertical component than control saccades (2D KS test; P = 0.02 to the right, P = 0.10 to the left). Once again, this trend was consistent across the population. The vectors in Fig. 6 express the difference in horizontal and vertical saccade amplitudes across vibration and control saccades, and the preferential upward distribution of these vectors emphasize that vibration saccades had a significantly greater vertical component (0.98 ± 0.67°; a 1-tailed t-test vs. a mean of 0, P < 10–10). The small horizontal shift leftward (i.e., ipsilateral to the side of vibration) also reached significance but again was of smaller magnitude than the vertical shift upward (–0.30 ± 0.51°; a 1-tailed t-test vs. a mean of 0, P < 10–6).
|
|
Finally, we also compared the radial direction of vibration and control saccades, analyzing only the subset of those saccades directed toward horizontal targets. For each experimental block, we derived the mean direction of both vibration and control saccades, and from these two means derived a rotation that expresses the difference between the mean direction of vibration and control saccades for each experimental session. Positive values for this measure imply that vibration saccades had a more elevated direction than horizontal saccades. The results of this analysis are shown in Fig. 7 and again confirm that vibration saccades were more elevated than control saccades (1-tailed t-test versus 0, P < 10–10; difference vector significant in 74/120 experimental sessions). Taken with the comparative analysis of saccade amplitudes and starting eye position, these directional results suggest that the shifts in spatial endpoints occur because saccade metrics are altered by neck muscle vibration.
|
Influence of neck muscle vibration on saccade latencies
We also analyzed the saccadic latencies (SRTs) of control and vibration saccades. From the representative example shown in Fig. 8, A and B, the SRTs for vibration saccades were significantly longer than for control saccades (Wilcoxon rank sum test: rightward saccades, P < 10–7, leftward saccades, P < 10–5). A strong, significant trend toward longer SRTs for vibration saccades was consistently observed across the population [Fig. 8C: Wilcoxon signed-rank test of medians: P < 10–10, 80 of 149 (54%) of median differences were significant].
|
Variations in parameters of vibrations
If the oculomotor effects of neck muscle vibration are mediated through modulations in neck muscle afferent information, these effects should vary systematically as the duration, amplitude, or frequency of vibration are altered. Accordingly, we completed a series of sessions in which one of vibration duration, amplitude, or frequency was varied between experimental blocks from the standard parameters (1,500 ms, 200 µm, 100 Hz) while the other parameters were kept constant. We will present the results of such variations in order, considering the effects of modulating vibration duration first.
VIBRATION DURATION. As expected, prolonging the duration of neck muscle vibration increased its influence on memory-guided saccades. Figure 9 A summarizes the shift in spatial endpoints for the sample of data points in which the duration was varied between 500, 1,500, and 2,500 ms between different experimental blocks. From Fig. 9A, it is obvious that the spatial endpoints of vibration saccades were shifted progressively more left-up for longer vibration durations relative to control saccades (both vertical and horizontal endpoint shifts were significant for all durations: 1-tailed t-test vs. 0, P < 10–5 for all vertical endpoint shifts; P < 0.01 for all horizontal endpoint shifts). To analyze this data, we performed a Kruskal-Wallis nonparameteric ANOVA (KW test) across vibration duration and used a post hoc Tukey-Kramer test (TK test) to test whether the measure at the longest vibration duration was greater than at the shortest vibration duration. Assessing the vectorial magnitude of the endpoint shift confirmed that it increased with increasing vibration duration (Fig. 9Bi; KW test, P < 10–4, TK test, P < 0.001). Similar shifts were observed in the horizontal and vertical components of the saccadic eye movements (data not shown), and the direction of vibration saccades was progressively shifted upward relative to control saccades for longer vibration durations (Fig. 9Bii; KW test, P = 0.01, TK test, P < 0.001). The comparative starting position of the eyes across control and vibration trials did not change with prolonged vibration durations (KW test, P = 0.26). These results confirm that the influence of vibration on saccade metrics increased for longer vibration durations.
|
; White et al. 1994), that both the horizontal and vertical variability of both control and vibration saccades increased significantly for longer delay durations (KW test, P < 0.0007 for all tests, TK test, P < 0.01). However, the comparative difference between vibration and control saccades only increased with prolonged durations in the vertical dimension and not the horizontal dimension (KW test, P = 0.58 for horizontal variability differences, P < 0.0001 for vertical variability differences, TK test, P < 0.05). Thus in the horizontal dimension, the increases in the variability for longer durations was the same for both vibration and control saccades, and it was only in the vertical dimension that the variability of vibration saccades increased more with longer durations than control saccades. We also examined ocular stability during the fixation interval, assessing both the differences in the number of microsaccades and the vertical eye velocity across vibration and control trials. Although the number of microsaccades during vibration trials increased with prolonged vibration durations (Fig. 9Biii; KW test, P < 10–5, TK test, P < 0.001), the difference in vertical eye velocity in vibration versus control saccades decreased with longer vibration durations (Fig. 9Biv; the effect of vibration was significant (KW test, P = 0.01, TK test between 500 and 1,500 ms durations, P < 0.05). This suggests that the increased upward drift of the eye induced by vibration occurs at a more or less fixed velocity, and the increasing number of microsaccades for longer vibration durations is only a consequence of the prolonged delay interval.
VIBRATION AMPLITUDE. Increasing the amplitude of vibration affected both the relative spatial endpoints as well as the ocular stability during the fixation interval (Fig. 10). The endpoints for vibration saccades were progressively shifted in a left-up direction for greater vibration amplitudes relative to control saccades (Fig. 10A; both vertical and horizontal endpoint shifts were significant across all amplitudes: 1-tailed t-test vs. 0, P < 10–5 for all vertical endpoint shifts; P < 0.05 for all horizontal endpoint shifts). The vectorial magnitude of the endpoint shift increased with greater vibration amplitudes (Fig. 10Bi, KW test, P < 10–4, TK test, P < 0.001), and as before, this change was related to increasing saccade components and upwardly deflected directions (Fig. 10Bii, KW test, P = 0.005, TK test, P < 0.05), and not to differences in the relative starting position of the eyes across different vibration amplitudes (KW test, P = 0.56). In contrast to increases in vibration duration, larger vibration amplitudes had no effect on either the horizontal or vertical variability of vibration saccades (KW test, P > 0.35). Increasing the vibration amplitude also increased both the relative number of microsaccades compared with control saccades during the fixation interval (Fig. 10Ciii; KW test, P = 0.001, TK test, P < 0.001) as well as increasing the difference in the upward eye drift (Fig. 10Biv; KW test, P = 0.007, TK test, P < 0.05). This latter effect contrasts to that which was observed with vibration duration, wherein the velocity of the upward eye drift did not change.
|
|
A final concern about the effects of neck muscle vibration is that they could be mediated by the vestibular organs, rather than neck muscle spindles, assuming that mechanical vibration conducts through the skull to the vestibular organs. To control for this possibility, we compared the influence of low-frequency (30 Hz) vibration that was presumably outside of the physiological range in which spindles indicate muscle lengthening to the standard parameters of vibration (100 Hz). Because we kept the amplitude and duration of vibration constant, we assumed that the mechanical transduction of any putative signal to the vestibular organs would be the same regardless of the vibration frequency. Over a number of experimental blocks, we consistently observed that the endpoint shifts in saccadic eye movements induced by neck muscle vibration at 30 Hz was negligible (Fig. 12 A; vertical and horizontal endpoint shifts were insignificant: 1-tailed t-test vs. 0, P > 0.35 for both horizontal and vertical endpoint shifts), even though vibration at a frequency of 100 Hz applied to the same location (the experiment was done in a blocked fashion) induced the now familiar left-upward shift in saccadic endpoints (Fig. 12B; both vertical and horizontal endpoint shifts were significant: 1-tailed t-test vs. 0, P < 10–5 for both).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Methodological considerations
Human studies of neck muscle vibration benefit from having the subject report the perceptual consequences of vibration, usually illusory motion of a fixated light, prior to beginning the experiments (Biguer et al. 1988; Karnath et al. 1993; Roll et al. 1991; Taylor and McCloskey 1991). Such a procedure optimizes the location of vibration in humans but is impractical in animals without extensive training on reporting directions of veridical motion. The inability to establish an appropriate location of vibration prior to an experimental session likely explains much of the variability in our results across experimental sessions (e.g., Fig. 3). Such variability is a common observation across different human subjects (Biguer et al. 1988; Han and Lennerstrand 1995; Karnath et al. 1994; Lewald et al. 1999; Taylor and McCloskey 1991).
Inherent to any study of animal behavior is the difficulty in ensuring the animal's cooperation, lest the results be attributable to task performance under aversive conditions. A number of observations suggest that our monkeys were cooperating with the task in both control and vibration trials. First, the animals quickly acclimatized to neck muscle vibration and began performing memory-guided saccades in the presence of neck muscle vibration on the first day it was introduced. Second, throughout all experimental sessions, we observed no tendency to perform control trials selectively at the expense of vibration trials; rather both monkeys performed both types of trials at an equally high rate. Third, there was no evidence for systematic differences in endpoint variability across control and vibration saccades as might have been expected if vibration caused the animals to be distracted from the task at hand.
Another difficulty in our study lies in inferring the spread of vibration, given the anatomical complexity of the head-neck musculoskeletal system. The dorsal neck musculature of both monkeys and humans is organized into four overlapping layers (Kamibayashi and Richmond 1998; Richmond et al. 2001), and it is not known how deeply surface vibration penetrates into these layers. This concern is aggravated by the plentiful endowment of muscle spindles in the deeper muscles closer to the spinal column (Cooper and Daniel 1963; Richmond and Abrahams 1975; Richmond and Bakker 1982). Although we cannot be sure whether the effects we have observed result from spindles residing in superficial or deep neck muscles, our demonstration that monkeys can serve as animal models for neck muscle vibration enables the development of implanted devices capable of delivering vibration selectively to certain muscle groups, or at the very least, quantify the spread of vibration to other muscles.
Finally, one possible reason for the relatively modest influences of neck muscle vibration observed in this study is the degree of conflict between the spindle activity entrained by vibration and other sensory modalities. Presuming that vibration does not affect the vestibular end-organs (this point is considered more thoroughly in the following section), any sense of head-re-body motion induced by vibration is countered by other senses which encode the absence of head motion in space (e.g., the vestibular end-organs) or the absence of body motion (e.g., cutaneous receptors; foot pressure receptors). Human studies have documented the contribution of somatosensory modalities to postural sensations (Lackner et al. 2000), and if we assume that similar capabilities exist in the monkeys, this likely leads to the perception of a lack of body motion within the chair. Furthermore, the monkeys in our study are presumably aware that their heads are firmly restrained to the chair, which again likely counters the perception of head-re-body rotation.
Given these many methodological concerns, it is perhaps surprising that neck muscle vibration in monkeys leads to any behavioral effects in the first place. However, these concerns apply equally to studies of neck muscle vibration in humans, yet numerous studies have documented perceptual and behavioral effects that demonstrate an important role for neck proprioception in the elaboration of the body schema for action (Biguer et al. 1988; Bove et al. 2002; Ivanenko et al. 1999; Karnath et al. 1994, 2002; Kavounoudias et al. 1999; Lewald et al. 1999; Popov et al. 1996; Roll et al. 1991; Taylor and McCloskey 1991). The development of an animal model of such effects enables more involved studies that can address many of these methodological concerns to lead to a more mechanistic understanding of the neural processing of neck muscle afferent information for action.
Evidence supporting causal role for neck muscle spindles
Numerous studies using a variety of methodologies have confirmed an important role for neck muscle afferent information in perception, postural control, and guided actions (e.g., Mergner et al. 1983, 1991, 2001; Nakamura and Bronstein 1995). However, given the uncertainty regarding the extent of neck muscle spindle activation via vibration, as well as the possible contributions of other sensory modalities, how certain are we that the effects of neck muscle vibration are mediated by altered discharge of neck muscle afferent information, and what are the alternative mechanisms? Barring direct measurements from muscle spindles or removal of other sensory modalities (e.g., via cutaneous anesthesia or vestibular lesions), we can only infer the contribution of these other modalities. However, considering the evidence from a diverse literature, our hypothesis of increased activity of neck muscle spindles is the most parsimonious explanation for our results.
We first consider the possibility that neck muscle vibration alters the output of vestibular end-organs. It has been known for some time that hair cells in both the vestibular end-organs are sensitive to mechanical vibrations of the skull (Young et al. 1977) and that skull vibration can lead to oculomotor effects similar to those described here (Lackner and Graybiel 1974). How possible is it that neck muscle vibration sets up pressure waves which propagate to and induce vibration of the skull? Such a question has not been completed resolved and is still a matter of some debate (Betts et al. 2000; Karlberg et al. 2002, 2003; McKenna et al. 2003).
While we cannot definitely exclude the possibility that the results we have observed are not, at least in part, secondary to stimulation of the vestibular end-organs, a number of observations suggest a predominant role for neck muscle spindles. Anecdotal reports suggest that the perceptual and behavioral effects of neck muscle vibration are abolished by xylocaine injections into neck muscle bellies (Lund 1980), even though this would not influence any purported skull vibration. Furthermore, there were no systematic differences between control and vibration trials using vibration frequencies of 30 Hz (Fig. 12). A frequency of 30 Hz likely lies outside of the physiological range indicating considerable muscle lengthening (Roll et al. 1989), but since vibration amplitude and duration were unchanged, such vibration should have propagated to the skull. Although the study by Young et al. (1977) studied vibration frequencies down to only 50 Hz, results in other species suggest that vibration frequencies of 30 Hz should excite at least a subset of vestibular end organs (Christensen-Dalsgaard and Nairns 1993; Hudspeth 1989).
There is also reason to suspect that neck muscle vibration may not induce skull vibrations in the first place. Recent animal and human studies (Freeman et al. 2000; Sohmer et al. 2000) show that skull vibration sets up fluid pressure waves, likely in the CSF, which can excite cochlear hair cells. Similar excitation can occur following vibration of the eye or cranial contents, even in the complete absence of skull vibration. Sohmer et al. (2000) also reported negligible auditory responses following neck muscle vibration, and although the location of vibration was left unspecified, this implies that pressure waves induced by neck muscle vibration are unlikely to induce skull vibration that subsequently transduced to the CSF. However, estimates of the propagation of vibration to the skull following vibration of the sternocleidomastoid muscle have yielded conflicted results (Betts et al. 2000; Bove et al. 2002), and it remains possible that the location of neck muscle vibration may be a prime determinant of the presence and magnitude of skull vibration.
These findings lead us to conclude that the increased afferent activity of neck muscle spindles, rather than vestibular end-organs, is the most likely route for the influence of neck muscle vibration, although we cannot completely discount possible vestibular influences. Information from cervical and vestibular sources converge at many sites throughout the neuraxis (Bottini et al. 2001; Gdowski and McCrea 2000; Grüsser et al. 1990; Hongo et al. 1988; Kasper et al. 1988), yet is conspicuously segregated in different areas of the posterior parietal cortex representing space in body-centered (area LIP) or world-centered coordinates (area 7a; Snyder et al. 1998). Compared with body-under-head rotation, the relatively selective activation of neck muscle afferents through localized mechanical vibration may be a useful methodology by which to study the origin, encoding and integration of head position signals within various brain areas.
Effects of neck muscle vibration on memory-guided saccades
Assuming that vibration alters neck muscle afferent information, how and why does this lead to the observed effects the oculomotor system? It seems likely that the ocular instability during the delay interval is due to the cervico-ocular reflex (COR). Neck muscle vibration in humans leads to nystagmus-like patterns of eye motion similar to what we observed here, and although the direction of eye motion in humans is predominantly horizontal, vertical, and torsional components have been reported, particularly in clinical populations (Han and Lennerstrand 1995; Karlberg et al. 2003; Popov et al. 1999; Strupp et al. 1998; Yagi and Ohyama 1996). That human subjects report illusory light motion in the direction opposite to the slow phase eye movements during vibration is particularly compelling evidence that such a perception is secondary to eye motion induced via the COR (Popov et al. 1999; Strupp et al. 1998). Given that we observed upward slow phase eye movements in both monkeys (Fig. 4), it is plausible that they perceived downwardly directed illusory motion of the FP during the delay interval. However, lacking an empirical report, this presumed percept remains speculative.
Regardless of whether the animals perceived illusory motion of the FP during the delay interval, the endpoints of memory-guided saccades were systematically shifted upward during vibration saccades. Such endpoint shifts are predominantly caused by saccades of altered metrics and are not simply a consequence of the accumulating upward drift during the delay interval (Figs. 5 and 6). Furthermore, the comparable endpoint variances across control and vibration trials suggest that the animals were generating saccades with the same precision. The upward shifts and prolonged SRTs of vibration saccades are also opposite to what would have been predicted if the downward microsaccades consequent to the COR were to interact with memory-guided saccades in a manner similar to that reported between VOR quick phases and antisaccades (Van Beuzekom and Van Gisbergen 2002).
Using the endpoints of memory-guided saccades as a proxy for the animals' estimates of target location, the animals apparently believe the target to be elevated relative to its veridical, space-fixed location. Our best explanation for this phenomenon is that the animals perceive that their gaze is deviated downward during the delay interval of vibration trials and compensate for this shift with a greater vertical component. The pattern of oculomotor responses with variations in vibration parameters suggests that the perception of downward gaze deviation accumulates during the delay interval depending on the integration of afferent input, because endpoint deviations were progressively larger for longer vibration durations, greater vibration amplitudes (consistent with greater penetration of vibration), and faster vibration frequencies (consistent with faster entrained spindle discharges).
This explanation leaves unanswered why altered afferent information from neck muscles should affect perceived gaze location to influence memory-guided saccades given that only retinocentric location of the remembered target relative to current eye position is theoretically required in a head-restrained preparation. However, both frontal and parietal oculomotor areas receive neck muscle afferent information (Barbas and Dubrovsky 1981a,b; Brotchie et al. 2003; Snyder et al. 1998) and the role of the superior colliculus extends beyond control of saccadic eye movements to fulfill a more general role in the control of eye-head gaze shifts (Corneil et al. 2002; Freedman and Sparks 1997; Klier et al. 2001). The pattern of neck muscle recruitment during isometric gaze shifts also changes as a function of initial head positions (Corneil et al. 2001). Our findings that memory-guided saccades in head-restrained preparations are influenced by neck muscle vibration are consistent with the CNS using neck muscle afferent information in the sensory-motor transformations for gaze shifts in general.
It is also unclear why SRTs were prolonged in vibration trials (Fig. 8). It is possible that the increased upward drift during the delay interval made the detection of FP offset more difficult, thereby prolonging SRTs. Alternatively, the sensory conflict between neck muscle afferent information and vestibular signals may lead to longer processing times prior to saccade generation.
Comparison to human results
To our knowledge, no one has examined the effects of neck muscle vibration specifically on memory-guided saccades in humans, although other experimental results are related. Most importantly, a number of groups have demonstrated the effects of neck muscle vibration on a variety of guided actions, including pointing (Biguer et al. 1988; Roll et al. 1991; Taylor and McCloskey 1991). The typical result is that open-loop movements are biased horizontally in the direction opposite to neck muscle vibration with variable vertical components. Furthermore, neck muscle vibration induces a predominantly horizontal COR-like movement, with the slow-phase directed toward the side of vibration, although varied vertical and torsional movements have also been reported (Han and Lennerstrand 1995; Popov et al. 1999; Strupp et al. 1998; Yagi and Ohyama 1996). Thus one apparent difference between humans and monkeys is that the effects of neck muscle vibration in humans are manifested usually in the horizontal dimension, yet in the vertical dimensions for monkeys.
We raise two possible explanations for this apparent species difference. Such differences may stem from the comparative size of the dorsal neck muscle mass in humans and monkeys. Dorsal neck muscle vibration likely influences at least two muscle groups in both humans and monkeys: splenius capitis and semispinalis capitis (semispinalis capitis has two compartments in monkeys: biventer cervicis and complexus; Kamibayashi and Richmond 1998; Richmond et al. 2001). In both species, splenius capitis turns the head horizontally, whereas semispinalis capitis acts to pitch the head up (Corneil et al. 2001; Vasavada et al. 2002). Thus one possible explanation for the species difference in the horizontal versus vertical influences of neck muscle vibration in humans and monkeys is the proportional differences in the extensor muscles masses. The extensor muscles in monkeys are proportionally larger and more complex, a trend which likely serves to stabilize the rostrally cantilevered head and neck against gravitational forces (Richmond et al. 1999a), particularly during the terrestrial mode of quadrupdeal locomotion preferred by rhesus monkeys (Napier and Napier 1985). We speculate that larger neck muscle masses are associated with a great per capita spindle content, which is consistent with the predominantly horizontal effects of neck muscle vibration in humans, and vertical effects in monkeys.
An alternative explanation of the differential effects in monkeys and humans lies in the location of neck muscle vibration. In our experiments, vibration was applied between 2 and 5 cm below the external occiput and 2 and 4 cm lateral of midline. In human experiments, experimenters typically optimize the location of neck muscle vibration for horizontal perceptions of self-motion or illusory light motion (Biguer et al. 1988; Karnath et al. 1993; Taylor and McCloskey 1991). The apparent species difference between humans and monkeys may therefore result simply from methodological differences in the location of vibration across different studies. Confirmation of a true species difference in the horizontal or vertical effects of neck muscle vibration will require experiments in both species which systematically map out the effects of different locations of vibration.
Conclusions and future directions
In closing, we believe we have successfully developed an animal model of the behavioral influences of neck muscle vibration. However, mechanistic study of neck muscle vibration will be hindered by the relatively small influences on the oculomotor system. In this study, standard parameters of vibration induced upward shifts in memory-guided saccades averaged about 1° for 10° target eccentricities (e.g., about 10%). Head position gain fields in area LIP, which measure the modulation of an oculocentric tuning curve by head-on-body position, range around 1.6%/° of change in head position (Snyder et al. 1998), thus the difference in delay period discharge in area LIP induced by neck muscle vibration will likely be only on the order of a few hertz. Furthermore, the existence of vertical head position gain fields has never been tested in area LIP for obvious technical reasons. As mentioned above, part of the reason for the relatively small influences of neck muscle vibration on the oculomotor system could be because of the conflicting messages issued by the vestibular system. A more fruitful approach may be to examine the influence of neck muscle vibration on open-loop pointing and reaching movements, where the degree of conflict between the vestibular and neck muscle afferent information will be lessened by the presumed changes in the position of the body under a stable head.
|
GRANTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACT |
|---|
