| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
The Journal of Neurophysiology Vol. 80 No. 5 November 1998, pp. 2391-2404
Copyright ©1998 by the American Physiological Society
1 Department of Neurology and the Center for Visual Science, University of Rochester, Rochester, New York 14642; and 2 Medical Research Council, Institute of Neurology, London, United Kingdom
 |
ABSTRACT |
|---|
 Abstract
 Introduction
Methods
Results
Discussion
References
|
|---|
Paige, Gary D., Laura Telford, Scott H. Seidman, and Graham R. Barnes. Human vestibuloocular reflex and its interactions with vision and fixation distance during linear and angular head movement. J. Neurophysiol. 80: 2391-2404, 1998. The vestibuloocular reflex (VOR) maintains visual image stability by generating eye movements that compensate for both angular (AVOR) and linear (LVOR) head movements, typically in concert with visual following mechanisms. The VORs are generally modulated by the "context" in which head movements are made. Three contextual influences on VOR performance were studied during passive head translations and rotations over a range of frequencies (0.5-4 Hz) that emphasized shifting dynamics in the VORs and visual following, primarily smooth pursuit. First, the dynamic characteristics of head movements themselves ("stimulus context") influence the VORs. Both the AVOR and LVOR operate with high-pass characteristics relative to a head velocity input, although the cutoff frequency of the AVOR (<0.1 Hz) is far below that of the LVOR (~1 Hz), and both perform well at high frequencies that exceed, but complement, the capabilities of smooth pursuit. Second, the LVOR and AVOR are modulated by fixation distance, implemented with a signal related to binocular vergence angle ("fixation context"). The effect was quantified by analyzing the response during each trial as a linear relationship between LVOR sensitivity (in deg/cm), or AVOR gain, and vergence (in m
1) to yield a slope (vergence influence) and an intercept (response at 0 vergence). Fixation distance (vergence) was modulated by presenting targets at different distances. The response slope rises with increasing frequency, but much more so for the LVOR than the AVOR, and reflects a positive relationship for all but the lowest stimulus frequencies in the AVOR. A third influence is the context of real and imagined targets on the VORs ("visual context"). This was studied in two ways
when targets were either earth-fixed to allow visual enhancement of the VOR or head-fixed to permit visual suppression. The VORs were assessed by extinguishing targets for brief periods while subjects continued to "fixate" them in darkness. The influences of real and imagined targets were most robust at lower frequencies, declining as stimulus frequency increased. The effects were nearly gone at 4 Hz. These properties were equivalent for the LVOR and AVOR and imply that the influences of real and imagined targets on the VORs generally follow low-pass and pursuit-like dynamics. The influence of imagined targets accounts for roughly one-third of the influence of real targets on the VORs at 0.5 Hz.
The vestibuloocular reflex (VOR) produces eye movements that compensate for head movements to maintain visual image stability. The VOR is compensatory for both angular (AVOR) and linear (LVOR) head movements, as required by the challenges of natural behavior. The VORs typically operate together with visual following mechanisms to achieve the common goal of preserving stable images of targets in space.
Subjects
Nine normal human volunteers between the ages of 21 and 49 participated in these experiments. Four participated in related experiments, and the remaining subjects were naïve. All were free of past or present neurological, ophthalmologic, otologic, systemic, or traumatic disease that could have affected results. All showed normal neuroophthalmologic and neurootologic function on clinical examination (performed by Paige), with particular attention toward assuring normal corrected visual acuity, stereopsis, visual fields, binocular oculomotor control, binaural hearing, and vestibular function. In addition, all subjects tested normally on pure-tone audiometry and vestibular caloric testing. The experimental procedures described were all performed in accordance with the 1964 Declaration of Helsinki and with the approval of an internal Research Subjects Review Board.
Eye movement recording and calibration
Horizontal and vertical eye movements were recorded binocularly with a dual charge-coupled device camera system (El-Mar, Toronto). The system consists of two cameras mounted on a goggle frame together with infrared (IR)-emitting diodes on both sides of each eye. The cameras view the eyes with the aid of half-silvered mirrors positioned below and in front of each eye but without impinging on the visual field necessary to view all targets in these experiments. The system tracks the motion of two IR reflections as well as the computed center of the pupil at 120 Hz and generates analog outputs proportional to horizontal and vertical eye position binocularly with a sensitivity of ~0.2° over a ±25° range. The method is established and compares well with others (DiScenna et al. 1995 Stimulus presentation and apparatus
CHAIR AND MOTION CONTROL.
A multiaxis sled/rotator (Contraves USA and JA Design, Pittsburgh, PA) was used to present all motion profiles (see Fig. 1). Subjects sat in a custom test chair with their heads securely fixed by a rigid bite bar of dental impression compound over a steel plate molded to each subject's bite. This tight interface between head and chair is as rigid as can be achieved practically and is essential if stimulus control is to be maintained during high-frequency motion. The chair with subject was mounted on a motorized linear sled riding on precision linear bearings and driven by a lead screw (1.1 m total excursion) connected to a DC motor by a timing belt. This mechanical drive was chosen over other methods because of its remarkable ability to faithfully generate high frequencies (including 4 Hz) of translation at high peak accelerations (
MECHANICAL CONSIDERATIONS.
Motion control in complex mechanical systems entails potential difficulties. How accurately does the motion profile of the head follow the desired control signal? The question can be parsed into two aspects, the mechanics of the sled/rotator device and the linkage between the device and the head. To assess the former, mechanical tests were performed for all stimulus parameters with a three-axis linear accelerometer (Entran EGAL3 12S-10D) and three-axis angular rate sensor (Watson Industries, Eau Claire, WI) mounted on the head-holder/bite-bar apparatus of the chair. Inaccuracies arose with increasing stimulus frequency (worst at 4 Hz). These were limited primarily to a lag in IA translation during linear trials caused by the sled's drive mechanism and a lag in yaw during rotation trials caused by chair compliance. Motion along or around axes outside the primary stimulus proved inconsequential. Correction factors for stimulus amplitude and phase were propagated throughout data analyses. These adjustments ensured an accurate depiction of motion at the bite bar.
CONDITIONS OF VVI.
Translational (LVOR) and rotational (AVOR) trials were studied under two conditions of VVI, 1) when targets were ef to allow visual enhancement of the VORs and 2) when targets were hf to permit visual suppression of the VORs. For each condition, targets were extinguished periodically to record the VORs in darkness. However, subjects were instructed to continue "fixating" the imagined hf (ihf) or ef (ief) targets. Thus trials were always paired, including ef/ief and hf/ihf forms, for both translational and rotational trial sets. This yielded a total of four trial types: ef/ief-LVOR, hf/ihf-LVOR, ef/ief-AVOR, and hf/ihf-AVOR. All stimulus characteristics were performed for each of these trial types.
Data collection and analysis
HARMONIC ANALYSIS.
All stimulus generation and data acquisition was performed by an IBM-compatible Pentium PC and custom software (Telford et al. 1997 GEOMETRIC CONSIDERATIONS.
Overview. The kinematic requirements for ocular responses to head rotation and translation can be directly calculated to provide a reference for comparison with actual VOR response measures. To maintain fixation on hf targets, no eye movement is required regardless of stimulus characteristics. In contrast, for ef targets, the ideal response differs between rotation and translation trials, according to the following kinematic requirements.
QUANTITATIVE ANALYSIS OF RESPONSE PERFORMANCE.
Data analysis proceeded in steps. Gains (AVOR) or sensitivities (LVOR) and phases for each stimulus condition and frequency were first distributed into vergence bins centered around the value required for each target distance. Vergences tended to cluster closely around these points even in darkness and only wandered on occasion in darkness after viewing the nearest targets, and then only in some subjects (see Fig. 2). Linear regressions were then performed on the data, yielding slopes (response parameters as a function of vergence) and intercepts (response at 0 MA vergence), in concert with the form of Eqs. 1 and 2. The number of cycles represented in each bin varied slightly as a function of frequency. Typically, 5-10 cycles were available at 0.5 Hz at each vergence and viewing condition, rising to 15-25 cycles at 4 Hz. Because vergence could change during a cycle, presumably modulating VOR performance accordingly, cycles were excluded if large and erratic changes in vergence occurred midcycle. This was rare, and in any case shifts in vergence presented little problem because the relationship between vergence and VOR response amplitude is linear (Busettini et al. 1994
MECHANICAL AND GEOMETRIC CONCERNS.
To what extent would the mechanical variables raised above (coupling of the motion stimulus and of the eye tracker to the head) affect experimental results? A key point is that all mechanical influences that we measured tend to increase with rising stimulus frequency, and all would produce fixed artifacts at any given frequency. In analysis, they would result in erroneous response intercepts. No such artifact would disturb the measured influence of visual targets or vergence on ocular responses, and these are the key parameters of interest in this report.
LVOR and VVI during head translation
LVOR RESPONSES RELATED TO FIXATION OF EARTH-FIXED TARGETS.
Examples of raw eye movement responses during head translation at 4 Hz are shown in Fig. 2. The stimulus (top trace) required to achieve 0.20-g peak acceleration at 4 Hz is quite small in excursion (see Table 1). A portion of the response record during an ef/ief trial from one subject is illustrated in the middle set of traces. Response amplitude varies considerably in relationship with vergence angle. Initially (far left in the diagram) the nearest target LED is on (thick bar above traces), and vergence is large (thick trace). The ocular response (thin oscillating eye position trace) is initially large in amplitude. This represents a portion of the trial under ef conditions. When the target is extinguished (at ~1.5 s), vergence gradually declines in darkness, as this particular subject is unable to maintain near fixation in the absence of visual feedback. The shift in vergence is accompanied by a gradual decline in response amplitude.
LVOR RESPONSES RELATED TO FIXATION OF HEAD-FIXED TARGETS.
Response characteristics were assessed in the presence of real and imagined head-fixed targets (hf and ihf). Examples of raw ocular responses are illustrated in Figs. 2 and 3 (lowest set of traces), corresponding to 4- and 0.5-Hz trials, respectively. At 4 Hz (Fig. 2), there is little apparent distinction between responses under hf and ef conditions or between ihf and ief conditions, so long as differences in vergence are taken into account. The presence of a near hf target simply cannot reduce the LVOR response driving the eyes. Subjectively, all subjects reported strong oscillopsia under these conditions. In contrast, all subjects were able to overcome the LVOR at 0.5 Hz (Fig. 3) and in general had little difficulty suppressing eye movements during either hf or ihf conditions. The LVOR was completely suppressed in the presence of real targets and was barely perceptible after targets were extinguished.
RESPONSE LINEARITY AS A FUNCTION OF STIMULUS INTENSITY.
In addition to quantifying LVOR-VVI response characteristics as a function of frequency, we assessed the influence of stimulus intensity at 4 Hz, where the LVOR is most robust. A subset of four subjects participated. Sinusoidal head translations were generated at 0.1-, 0.2-, 0.3-, and 0.4-g peak head acceleration over all conditions of VVI (ef, ief, hf, and ihf) and all target distances. Analysis proceeded as described previously. Linear regressions revealed no significant influence of stimulus amplitude on response parameters (slope, intercept, or phase). Most notably, sensitivity slope varied by <8% across the range of stimulus amplitude for all conditions and in no systematic or significant manner.
AVOR and VVI during head rotation
AVOR RESPONSES RELATED TO FIXATION OF EARTH-FIXED TARGETS.
The entire experiment described previously was repeated with head rotation as the stimulus instead of head translation. All attributes of VVI condition (ef, ief, hf, and ihf) and target distance were identical, and the same subjects were used. A key element of this part of the study is that AVOR-VVI data were processed exactly as were LVOR-VVI data, including the analysis of the influences of binocular vergence, stimulus frequency, and VVI condition. Consider the raw eye velocity response to 4-Hz rotation shown in Fig. 6 (middle record). The subject is initially fixating the middle target and maintains vergence quite well after the target is extinguished (at ~1 s). The response amplitude is likewise maintained. After the near target is presented at ~4 s, vergence suddenly climbs, accompanied by an obvious increment in response amplitude. This persists after the target is once again extinguished, and gradually declines as vergence tapers off beyond 8 s. These qualitative characteristics resemble comparable responses seen during head translations (Fig. 2), but the modulation by vergence is less pronounced during rotation.
AVOR RESPONSES RELATED TO FIXATION OF HEAD-FIXED TARGETS.
Response characteristics were assessed in the context of hf as well as ef ones. Examples of raw ocular responses are illustrated in Figs. 6 and 7 (lowest set of traces), corresponding to 4- and 0.5-Hz trials, respectively. At 4 Hz (Fig. 6), hf and ihf responses are robust and barely distinguishable from ef and ief conditions. As during translation trials, the presence of a near hf target cannot reduce the vestibular drive to the eyes, and all subjects again reported strong oscillopsia. All subjects were able to overcome the AVOR at 0.5 Hz (Fig. 7) and had little difficulty suppressing most (but not all) of the eye movement response during fixation of an hf target. However, a strong AVOR response appeared promptly when the target was extinguished but still imagined (ihf), although at smaller amplitude than under ief conditions. This is in contrast to the barely visible response seen under comparable conditions during translation trials (Fig. 3, bottom traces).
Overview
The response dynamics of the VOR and its interactions with vision were quantified during both translational and rotational motion (LVOR-VVI and AVOR-VVI). Response characteristics were assessed as a function of three variables known to influence the VORs, all related to the context in which head movements are made. These contextual influences include stimulus characteristics (frequency and amplitude of motion: stimulus context), binocular viewing distance (fixation context), and real or imagined visual target motion (visual context). Stimulus properties directly affect the vestibular input by virtue of both the mechanics of the sensory endorgan and neural processing within VOR pathways. In contrast, the fixation and visual contexts reflect the presumed goal of all VORs, to maintain binocular fixation on targets in space. This goal is determined by the geometric relationship between the position and motion of the eyes relative to their fixated target as well as the potential motion of the target in space. These relationships govern the ocular responses required to maintain stable binocular foveal images during head movements. In this study, common analytic methods were applied to LVOR-VVI and AVOR-VVI responses in part to emphasize their common goals.
Influence of stimulus characteristics on VVI
Ocular responses during translational and rotational trials were quantified over the frequency bandwidth, 0.5-4.0 Hz. This limited range was chosen largely because previous experiments in monkeys (Paige and Tomko 1991b Influence of binocular viewing distance on VVI
Simple geometric considerations dictate that the ideal VOR must modulate its response amplitude inversely with viewing distance if binocular fixation stability is to be maintained. This is particularly true for the LVOR, in which case the kinematically ideal response to head translation should be entirely dependent on fixation distance. The ideal LVOR sensitivity should equal 0 when fixation distance is infinitely far and should climb progressively as fixation distance declines. This relationship is conveniently quantified by the sensitivity slope parameter (Fig. 5), which relates LVOR response amplitude to binocular vergence, a measure of the eyes' fixation distance. This relationship indeed holds for the experimentally recorded LVOR. However, the phenomenon is most robust at 4 Hz and declines as stimulus frequency drops. In other words, the influence of vergence on the LVOR is frequency dependent and behaves with high-pass dynamics (Telford et al. 1997 Influence of real and imagined visual targets on VVI
The presence of real or imagined visual targets has been known to influence AVOR response properties for decades (Barnes 1993
INTRODUCTION
Abstract
Introduction
Methods
Results
Discussion
References
). The lower cutoff frequency of the AVOR is <0.1 Hz, whereas that of the LVOR is ~1 Hz. Both perform well at high frequencies that exceed the capabilities of visual following. The fastest continuous visual tracking system in primates is smooth pursuit. Its low-pass performance characteristics allow accurate target tracking that is limited to relatively low frequencies, below~2 Hz. Thus visual-vestibular interactions (VVI) operate in complementary fashion to maintain target fixation across a broad range of stimulus frequencies, with each modality contributing most where the other fails.
eye movements serve to stabilize binocular fixation on targets in space and therefore maintain a stable bifoveate image. The geometry of eye and head movements in relation to binocular fixation of targets then becomes critical for understanding both the goals and performance characteristics of the reflex. For a given head movement, the VOR must compensate for both its rotational and translational components to maintain binocular target fixation. Proper compensation for the translational component is governed by target distance, or more accurately, fixation distance (how far away the 2 eyes are looking). Because the eyes are typically looking at a target, fixation and target distance are usually the same. Restricting attention to horizontal LVOR responses during interaural (IA) linear head motion, the ideal compensatory response is inversely proportional to fixation distance. If fixating a target far away, little or no eye movement is required, but, as fixation distance declines, ocular responses must progressively increase in amplitude. The LVOR indeed operates in this fashion, although imperfectly (Baloh et al. 1988; Paige 1989; Paige and Tomko 1991b; Schwarz and Miles 1991; Skipper and Barnes 1989). The source of fixation distance information underlying LVOR modulation remains controversial. Experimental manipulations of binocular vergence angle and accommodation (focus) of the intraocular lens reveal that vergence angle is the key variable related to VOR amplitude modulation in humans (Hine and Thorn 1987; Paige 1989, 1991), and accommodation plays a parallel but less robust role in rhesus monkeys (Schwarz and Miles 1991). It is not vergence angle per se but rather an as-yet unidentified "vergence command" signal that modulates the LVOR along with generating binocular vergence of the eyes because LVOR changes occur in advance of binocular vergence movements (Paige 1991; Paige and Tomko 1991b; Snyder et al. 1992).
; Paige 1996; Snyder and King 1992; Telford et al. 1998; Viirre et al. 1986). However, studies employing modest frequencies of head rotation reported meager or even the opposite effect on the VOR (Crane et al. 1997; Gizzi and Harper 1996; Shelhamer et al. 1995; Viirre and Demer 1996) in which gain declines rather than increases with decreasing fixation distance. This study reevaluated the controversy in light of our recent finding that the influence of fixation distance on the monkey LVOR behaves with high-pass characteristics governed by a cutoff frequency >1 Hz (Telford et al. 1997). Perhaps the same applies to the AVOR.
; Furst et al. 1987). Recall that the VORs generally operate in conjunction with visual mechanisms that also drive eye movements, primarily the smooth pursuit system given the range of stimulus frequency considered here (0.05-4.0 Hz). VVI is typically studied in two ways, when the visual target is earth-fixed (ef) or moving in tandem with the head (head fixed, hf). In the former case, visual input enhances the VOR, and, in the latter case it suppresses the reflex, regardless of the nature of the head movement (i.e., rotation or translation). A fascinating feature of VVI is the apparent modulation of VOR responses in darkness by the context of imagined target motion relative to the head. That is, simply imagining that a "target" in darkness is ef or hf (changing visual context) results in a VOR enhancement or attenuation, respectively. The influences of real and imagined targets on the VOR behave with similar dynamic limitations that resemble the low-pass (<2 Hz) characteristics of smooth pursuit (Barnes 1993; Collewijn 1985; Paige 1994). Although the LVOR has been studied under different conditions of VVI and visual context (Baloh et al. 1988; Israël and Berthoz 1989; Oas et al. 1992; Shelhamer et al. 1995; Skipper and Barnes 1989), none measured fixation distance while also addressing LVOR behavior in the critical high-frequency range of head movements where the reflex is most robust.
).
METHODS
Abstract
Introduction
Methods
Results
Discussion
References
; Paige et al. 1996).
; Furst et al. 1987; Kasteel-van Linge and Maas 1990; Paige 1994). Mean scale factors derived from these trials were then used to adjust responses from other trials. Although refractive correction cannot be worn during recordings, refractive blur does not affect oculomotor performance tasks appreciably (Post et al. 1979; Van den Berg and Collewijn 1986), and our experience is in concert with this contention.
0.5 g), despite a~400-lb. payload. The sled, chair, and subject were in turn mounted on a motorized angular base axis (peak torque, 325 ft-lb.) that performed all rotational trials.
View larger version (31K):
[in a new window]
FIG. 1.
Schematic of the sled/rotator device with a subject positioned for interaural (IA) translation or earth-horizontal rotation (R). The geometric requirements for an ocular response (angle) to translation (T) while the subject fixates a target (apple) at distance, d, is illustrated here and described further in the text.
View this table:
TABLE 1.
Stimulus parameters
, 1998). Binocular horizontal and vertical eye position, chair angular velocity, sled linear velocity, and target ON/OFF signals were analog filtered (50 Hz) and sampled at 100 Hz. During analysis individual cycles were identified with a timing record produced by the stimulus generator within the software. At 2.0 and 4.0 Hz, a large number of cycles were available that were saccade-free, regardless of fixation conditions (e.g., ef vs. hf). These were selected for analysis, whereas those containing saccades were discarded. Analysis was performed on position traces at these high frequencies to avoid digitization noise, especially given the small response amplitudes recorded. At lower frequencies, eye position signals were first digitally differentiated and smoothed to yield eye velocities. Saccades were more frequent, particularly during hf/ihf-AVOR and ef/ief-LVOR trials. These saccades were identified and removed from eye velocity records with an iterative least-squares sinusoidal fit and windowing procedure (Ebisawa et al. 1988; Paige and Sargent 1991a), replacing gaps with the corresponding portions of the sinusoid fit to each cycle. An initial fit "by eye" was performed on each cycle, with a 50°/s inclusion window around the fit. Subsequent fits (the fundamentals from Fourier analysis) and window reductions (scaled by residual root-mean-square noise) for each cycle rapidly converged on stable desaccaded cycles after two to three iterations. Saccades comprised a small proportion of each cycle and were removed only from the 0.5- and 1-Hz trials, as noted previously.
1). For example, one MA would be required to fixate a target 1 m away, whereas two MAs would be required for a target 0.5 m away. This unit provides a form of vergence normalization that allows direct comparisons between different subjects and species regardless of head size and ocular separation (6.4 cm on average in humans).
; Telford et al. 1997). Figure 1 illustrates the geometry of compensatory LVOR responses to IA head translation. For small angles, the ideal LVOR response (
i) is determined by the magnitude of the head translation (T) and the target distance (d) according to the relation
where
(1)
is vergence (in MA) and Si is sensitivity (in deg/cm).
; Paige 1991; Paige and Tomko 1991b; Schwarz and Miles 1991; Telford et al. 1997), and this forms the basis of our quantitative assessment of LVOR performance.
; Hine and Thorn 1987; Paige 1996; Viirre et al. 1986). The magnitude of the ocular response required to compensate for this translational component depends on fixation distance. When fixating far away, the required AVOR gain remains unity, but increasingly larger gains are needed as fixation distance declines. The ideal gain, Gi, required to maintain binocular fixation is a function of both ocular eccentricity from the axis of rotation, r (in m), and fixation distance as determined by vergence angle, (in MA). This geometric relationship between gain, vergence, and ocular eccentricity (Telford et al. 1998) can be approximated for small angles of rotation by the equation
During natural (head-centered) head rotation, r = 0.08 m in humans, but this distance was in practice extended by ~0.03 m because of the mechanics of the bite-bar apparatus used to hold the head.
(2)
; Paige and Tomko 1991b; Schwarz and Miles 1991; Telford et al. 1997). Within-cycle shifts in vergence are matched by shifts in response amplitude within the cycle, and both parameters are derived from the entire cycle.
View larger version (38K):
[in a new window]
FIG. 2.
Raw records of horizontal eye movement responses (middle and bottom sets of traces; all digitally smoothed) to translation at 4 Hz (oscillating trace in top set) from one subject while attempting to fixate a visible earth-fixed (ef, middle records) or head-fixed (hf, bottom records) target (thick bar above response records indicates that the target is on) and when instructed to maintain fixation on imagined targets in darkness. The responses (thin oscillating traces; axes labeled on the left) display evidence of modulation associated with changes in vergence (thick traces; axis labeled on the right), plotted in meter angles (MA, in m 1). Because all targets were roughly aligned with the right eye, vergence was generated almost entirely by the left eye, and the average position of the right eye remained relatively stable.
; Paige and Tomko 1991b; Snyder et al. 1992), and perhaps this factor might distort results. Quantification of the precise latency difference has proven problematic, but the difference averaged only 49 ms in the study by Snyder et al (1992). Any influence of this small difference would appear in the noise (SDs) of averaged binned data presented and plotted. Errors caused by latency differences between vergence and VOR responses are small and cannot account for the robust interaction observed between them. In general, the results of all regressions proved precise regardless of VVI condition and frequency; the SE of the coefficients (slope and intercept) typically remained <0.02 and only rarely (and idiosyncratically) exceeded 0.03.
). Again, this concern is inconsequential for all but the largest excursions and the closest ef target. Ideal vergence during head oscillation varies during the cycle at twice the stimulus frequency and with an average over the cycle (recall that the average is used in analysis) that is 10% less than the 5.0 MA required for the initial target distance of 20 cm. This error drops precipitously as peak head translation or vergence declines. The phenomenon appears in some raw vergence records, especially those in Fig. 3 (ef/ief-LVOR at 0.5 Hz). Small oscillations in vergence records at 4 Hz (e.g., Fig. 2) also include a component due to small unavoidable differences in calibration between the right and left eye movement signals. This is especially true when vergence is large, and therefore requires large adductions of the left eye while the right eye remains roughly fixed (recall that targets were roughly aligned with the right eye).
View larger version (31K):
[in a new window]
FIG. 3.
Raw records of horizontal eye movement responses to translation at 0.5 Hz, displayed as in Fig. 2.
RESULTS
Abstract
Introduction
Methods
Results
Discussion
References
View larger version (20K):
[in a new window]
FIG. 4.
Linear vestibuloocular relfex (LVOR) response sensitivity (deg/cm), plotted as a function of vergence on a cycle-by-cycle basis, at both 4 Hz (top plot) and 0.5 Hz (bottom plot). Open circles depict when the ef target was on, and closed circles indicate when the target was off (imagined fixation in darkness, or ief condition). The thick solid lines illustrate ideal response properties based on geometric predictions, and the dotted (ef) and dashed (ief) lines show best-fit lines to the data. Note the markedly reduced slope of the ief slope at 0.5 Hz compared with 4 Hz and the presence of a small response sensitivity at 0 MA vergence (response intercept) in all but the ef condition at 0.5 Hz.
; Paige and Tomko 1991b; Telford et al. 1997). Linear regressions provided a convenient means of characterizing response properties. The solid lines in Fig. 4 represent the geometrically ideal relationship, calculated to have a slope of 0.57°/cm/MA and an intercept of 0°/cm at 0 MA of vergence, regardless of frequency. The dotted and broken lines show the regression results for the ef and ief conditions, respectively. Several observations are derived from these plots. First, there is a substantial positive intercept under both ef and ief conditions at 4 Hz. Second, the sensitivity slope for the ef condition is greater than that for ief and closer to the ideal at both frequencies. Third, the sensitivity slope in darkness (ief) at 0.5 Hz is considerably reduced, and farther from ideal, than that at 4 Hz. This last observation is apparent in the raw records of Figs. 2 and 3 and illustrates the failure of the LVOR to adequately drive smooth eye movements at even modest frequencies. This was true despite the context of an imagined ef target. That this context actually existed during responses is apparent in Fig. 3, as the subject employed saccades to supplement the LVOR and more closely approximate the goal of maintaining fixation stability during translation (Israël and Berthoz 1989).
View larger version (28K):
[in a new window]
FIG. 5.
Mean LVOR sensitivity slope (response sensitivity as a function of vergence), intercept (response sensitivity at 0 vergence), and phase, plotted as a function of stimulus frequency for all conditions of target motion (see key). Error bars depict SDs. The ideal response for ef/ief conditions would be a slope of 0.57°/cm/MA, an intercept of 0°/cm, and a phase angle of 0°. For hf/ihf conditions, the ideal slope would also be zero.
View larger version (65K):
[in a new window]
FIG. 6.
Raw records of horizontal eye movement responses to head rotation at 4 Hz, displayed as in Fig. 2 and from the same subject.
View larger version (29K):
[in a new window]
FIG. 7.
Raw records of horizontal eye movement responses to head rotation 0.5 Hz, displayed as in Fig. 2.
2° of 0 for both ief and ef conditions, except at 2 and 4 Hz, where ef phase developed a lag of -3 ± 3° and -5 ± 3°, respectively (P < 0.01).
View larger version (25K):
[in a new window]
FIG. 8.
Mean angular vestibuloocular reflex (AVOR) gain slope (response gain as a function of vergence), intercept (response gain at 0 vergence), and phase, plotted as a function of stimulus frequency for all conditions of target motion (see key). Error bars depict SDs. The ideal response for ef/ief conditions has a slope of 0.11/MA, an intercept of 1.0, and a phase angle of 0°. For hf/ihf conditions, the ideal slope and intercept are 0.
DISCUSSION
Abstract
Introduction
Methods
Results
Discussion
References
to what extent do the experimental conditions studied here reflect the physiological challenges of natural behavior? First, the VORs were studied in the presence of visual targets and therefore as interactions with visual mechanisms that also drive eye movements, most notably smooth pursuit. Natural activities typically entail such interactions, and objects of interest are encountered that are stationary (ef) or moving through space. The latter might resemble the experimental hf context, arising when examining objects in our hands while walking or when tracking targets largely with the head. Second, we studied the VORs during sinusoidal motion. Even casual observation reveals that harmonic oscillation is a common attribute of daily activity. Examples include head movements made during natural ambulation (Cappozzo 1981; Demer and Viirre 1996; Grossman et al 1988; Waters et al. 1973), which predominantly lie within the frequency bandwidth (0.5-4 Hz) employed in this study. Third, most behaviors include combinations of linear and angular motion, with different weightings depending on the activity (e.g., walking entails largely linear head motion). Extensive studies of VOR responses during angular, linear, and combined stimulation in squirrel monkeys revealed that AVOR-LVOR interactions behave linearly (Telfrod et al. 1996, 1998). The current human investigation limited stimuli to angular and linear motion alone to focus on specific contextual influences on the two classes of VOR. Preliminary (and unpublished) observations suggest that human AVOR-LVOR interactions also behave linearly, as supported by others (Crane et al. 1997).
; Telford et al. 1997) demonstrated a robust LVOR only during oscillations at >1 Hz or during transient accelerations that resemble high frequencies (Busettini et al. 1994; Schwarz and Miles 1991) but weak responses at more modest frequencies (Telford et al. 1997). In general, the current experiments in humans reflect the same overall characteristics. LVOR responses in darkness (ief and ihf conditions) generally display the greatest sensitivities and smallest phase leads at the highest frequency and show declining sensitivities and rising phase leads as frequency declines toward 0.5 Hz, although with clear influences of other contextual variables. In general, this qualitative behavior replicates findings in squirrel monkeys under similar conditions (Telford et al. 1997) and is characteristic of a high-pass filtering process operating on otolith input. The findings are consistent with the notion that the LVOR operates with distinct frequency-dependent reflexes that include a high-frequency translational LVOR that operates at frequencies >1 Hz and a tilt LVOR that operates most prominently below 0.1 Hz (Paige 1996; Paige and Tomko 1991a). Because our goal was to emphasize translational LVOR reflexes and their interactions with vision, a stimulus bandwidth was chosen that emphasized this reflex. A related frequency parsing scheme seems to hold, at least in part, for motion perception (Glasauer 1995; Guedry 1974; Schöne and Mortag 1968; Walsh 1961). A relevant psychophysical observation is that all subjects in these experiments reported subjective sensations of IA translation and never roll tilt. This is in contrast to low-frequency (0.1 Hz) linear accelerations in our laboratory with the same device (Seidman and Paige 1996; Seidman et al. 1998), which induced sensations of tilt and not translation. Studies at intermediate frequencies induce combinations of tilt and translation perceptions, commonly referred to as the "Hilltop illusion" (Glasauer 1995).
) by roughly a decade. Further, the low-frequency range of the AVOR is an extended version of the dynamics of its canal afferent input, accomplished by a process commonly called "velocity storage" (Raphan et al. 1979; Robinson 1981), whereas the translational LVOR specifically filters out low-frequency information from its otolith afferent input. This distinction is quite useful during natural behavior and serves to avoid erroneous LVOR responses and perceptions of translation when we simply tilt our heads toward one shoulder, a maneuver that activates IA otolith input as readily as translational accelerations (Paige et al. 1995; Paige and Tomko 1991a). That natural head tilt is registered by a robust canal signal in addition to otolith input does not alter the fact that when the head tilt is completed the canal signal ceases while otolith signals persist, and yet no enduring translational LVOR or perception accompanies this linear acceleration signal. This is why our model of the translational LVOR includes an effective two-pole, high-pass process within its central pathway (Telford et al. 1997).
). This observation applies to the LVOR in darkness but with a further modulation by the imagined context of target motion; that is, there is an obvious distinction between ief and ihf conditions at lower frequencies. Nevertheless, for both conditions, a clear influence of vergence is apparent at a fixed high frequency and a clear influence of stimulus frequency at a given vergence, as reported in monkeys (Telford et al. 1997). The high-pass properties of the vergence influence helps explain why many studies, typically employing lower frequencies or "soft" transients, have not witnessed a robust effect, and this in turn emphasizes the need for high-frequency stimuli in quantifying the LVOR.
; Paige 1991; Paige and Tomko 1991b; Schwarz and Miles 1991), including this one, is the presence of a response at zero vergence, as quantified by the sensitivity intercept parameter. This response when binocular fixation distance is infinitely far is clearly not required to maintain stable fixation. Indeed, any response under these conditions actually generates retinal image slip. We previously suggested (Paige and Tomko 1991b) that this positive intercept serves a useful purpose. In effect, the intercept shifts the entire relationship between vergence and response amplitude upward. This positive intercept allows the LVOR to approach ideal performance over a wider range of fixation conditions than if the intercept were zero. This concept is exemplified in Fig. 4, which includes the best-fit line segment (slope and intercept) to the data along with the ideal (slope of 0.57°/cm/MA and intercept of 0°/cm). In this subject, note that in the ief case at 4 Hz the regression line intersects the ideal at~3 MA of vergence, indicating a fixation distance of 33 cm. This intersection corresponds to perfect reflex performance; that is, the LVOR response exactly matches requirements to maintain binocular fixation on a point 33 cm from the eyes. For this particular subject, the ef-LVOR response at 4 Hz lies entirely above the ideal across the measured range of vergences but is idiosyncratic to this individual. On average, the point of perfect fixation stability occurs somewhere between 33 and 73 cm at 4 Hz across all conditions, including real or imagined ef and hf targets. Thus the LVOR seems optimized to maintain fixation stability at roughly arm's length. In addition, the error that accumulates as vergence extends from this distance in either direction remains quite small. Note that if the intercept were zero ideal performance would exist only when fixating infinitely far targets and would progressively deviate below the ideal as vergence increased. This would be especially troublesome at 4 Hz because the sensitivity slope is just 61% of ideal even when the LVOR is visually enhanced by a real ef target, and just 46% in darkness. Curiously, the LVOR performance in darkness observed in this study is considerably better than previously reported in the human LVOR (Busettini et al. 1994; Paige 1989, 1991). Again, the difference is likely due to the remarkable frequency dependence of the LVOR, including its vergence influence. Even a shift from 3 to 4 Hz provides a noticeable augmentation of LVOR response sensitivity.
; Paige and Tomko 1991b; Telford et al. 1997). Recent recordings of the human (unpublished observations) and monkey (Paige et al. 1995) translational LVOR at frequencies <0.5 Hz confirm this supposition.
; Snyder and King 1992; Viirre et al. 1986), the slope declined systematically as frequency decreased, to yield negative numbers at the lowest frequencies. This latter phenomenon was reported by others (Crane et al. 1997; Shelhamer et al. 1995; Viirre and Demer 1996). Thus AVOR gain was modulated by vergence in different directions depending on the frequency of the stimulus. At low frequency vergence had a detrimental effect on AVOR gain, whereas at the highest frequency AVOR amplitude climbed slightly with rising vergence. The overall increasing trend in gain slope with increasing frequency resembles a comparable finding in human LVOR-VVI responses and in the monkey LVOR (Telford et al. 1997).
; Maxwell and King 1992). The outcome would presumably be reflected as a balanced increase in muscle torque on the eye, without a change in its ambient eye position. This effective co-contraction would produce an increase in stiffness and/or resistance, which in turn might result in smaller response amplitudes for the same stimulus. Thus, as vergence increases and mechanical stiffness or resistance rises, response amplitudes presumably decline, resulting in a negative gain slope. Such a change in plant dynamics would be expected to behave in a frequency-dependent manner; specifically, high frequencies would be less influenced than low, as observed experimentally. Why is gain slope negative at low frequencies in the AVOR while sensitivity slope is always positive in the LVOR? Presumably, vergence-dependent changes in ocular mechanics would hold regardless of the particular vestibular reflex, but, because the LVOR is more dramatically influenced by vergence, the effect is presumably masked, resulting in a generally positive slope, although always suboptimal.
; Telford et al. 1996, 1998). In general, the outcome reduced the gain slope by ~0.01 across the frequency bandwidth and the intercept by only 2% at 4 Hz, declining to negligible values at the lowest frequency.
; Barr et al. 1976; Furst et al. 1987; Jell et al. 1988). AVOR-VVI responses reported here replicate earlier findings, but response properties are expressed in a nontraditional manner. Gain is parsed into an angular response component at 0 vergence (gain intercept) and a vergence-dependent response component (gain slope) that compensates for ocular translations in the head during natural rotation. Nevertheless, conclusions remain the same. The influence of hf and ef real visual targets is always greater than their imaginary equivalents. The effect is generally most pronounced at low frequency and declines systematically as frequency rises.
; Jell et al. 1988; Paige 1994). The same properties and conclusions apply to LVOR-VVI responses. The major difference between AVOR and LVOR driven responses is the generally more robust AVOR than LVOR at lower frequencies. This is due to the vastly different high-pass dynamics of the two reflexes. One result is that AVOR responses to rotation under ief conditions closely follow those under ef conditions at all frequencies, whereas LVOR responses to translation under ief and ef conditions are close only at high frequency and separate as frequency declines.
; Israël and Berthoz; Skipper and Barnes 1989). The smooth component of the response, presumably related to vestibular (LVOR) input, is meager at best. The same applies during ief-AVOR trials, but because the AVOR is so much more robust than the LVOR at, for example, 0.5 Hz, there is little need to invoke saccades along with the AVOR to maintain fixation stability; the AVOR performs sufficiently well that the overall response remains smooth. The situation reverses when considering ihf conditions for the same 0.5-Hz head oscillation. Whereas the LVOR requires little suppression to dampen its weak output at this low frequency, the more robust AVOR persistently drives the eyes away from the imagined hf target, and subjects must evoke saccades to repeatedly correct for the VOR-driven fixation errors. None of this occurs at 4 Hz, where any saccadic strategy would prove counterproductive given the small ocular excursions involved and the unacceptably long latency of saccade production relative to the 250-ms period of head oscillation.
) and monkeys (Cullen et al. 1991), but not during sinusoids (Paige 1994), as replicated here at 4 Hz. The same is true in the LVOR. The reason remains mysterious. Potentially, steady-state sinusoids reduce suppression of the VOR at high frequency by degrading fixation. The VOR never quite matches kinematic requirements at 4 Hz and therefore experiences persistent image slip under steady-state conditions. Studies with transient stimuli never operate under the same conditions and therefore represent a more idealized situation where fixation and image stability are initially quite well preserved before the head is perturbed. Perhaps a transient applied during high-frequency oscillation would provoke no greater change in VVI response properties than the steady-state response alone.
; Paige et al. 1996). Similarities extend to other response attributes, such as eye velocity oscillations (ringing) during both pursuit and hf-AVOR, and the phenomenon of prediction (Barnes 1993). Perhaps smooth pursuit really represents a multisensory tracking system for which vision is the primary, but not the only, driving stimulus to the system. If so, to what extent are vestibular signals available to this mechanism? We can use LVOR-VVI and AVOR-VVI recordings at 0.5 Hz to generate a reasonable estimate. The difference in peak eye velocity between ef and hf responses reflects the direct visual influence on eye movements (smooth pursuit) during both translational and rotational motion. Similarly, the difference between ief and ihf reflects the influence of imagined target motion (pseudo-pursuit). The ratio of the two differences [(imagined/(real)] provides an estimate of the desired answer. This ratio was calculated for both translational and rotational trials at 0.5 Hz and over a range of vergence angles between 0 and 5 MA. The ratios for angular and translational trials closely overlapped as a function of peak eye velocity and were therefore pooled. The ratio dropped systematically with increasing peak eye velocity from roughly 45% at 20°/s to 31% at 85°/s. Such high eye velocities were achieved during translational trials when subjects viewed or imagined the nearest target. Overall, and acknowledging the amplitude-dependent nonlinearity, the influence of imagined target motion (pseudo-pursuit) on the VOR generally accounts for roughly one-third of the influence of real targets (true smooth pursuit).
| |
ACKNOWLEDGEMENTS |
|---|
The authors thank P. Boulos, J. Cai, C. Chen, M. Gira, and K. Martin-Smith for technical and clerical assistance.
This research was supported by National Institutes of Health Grants AG-06442, 04935, RR-06853, and T32-EY-07125 to the Center for Visual Science. G. D. Paige was also supported by a grant from Research to Prevent Blindness.
| |
FOOTNOTES |
|---|
Address for reprint requests: G. D. Paige, Dept. of Neurology, Box 605, University of Rochester, 601 Elmwood Ave., Rochester, NY 14642.
Received 27 February 1998; accepted in final form 13 July 1998.
| |
REFERENCES |
|---|
|
Abstract
Introduction
Methods
Results
Discussion
References
|
|---|
This article has been cited by other articles:
|
|
 |
|
  A. A. Rader, C. M. Oman, and D. M. Merfeld Motion Perception During Variable-Radius Swing Motion in Darkness J Neurophysiol, October 1, 2009; 102(4): 2232 - 2244. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. Zhou, Y. Xu, I. Simpson, and Y. Cai Multiplicative Computation in the Vestibulo-Ocular Reflex (VOR) J Neurophysiol, April 1, 2007; 97(4): 2780 - 2789. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. M. Green, H. Meng, and D. E. Angelaki A Reevaluation of the Inverse Dynamic Model for Eye Movements J. Neurosci., February 7, 2007; 27(6): 1346 - 1355. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. A. Yong, G. D. Paige, and S. H. Seidman Multiple Sensory Cues Underlying the Perception of Translation and Path J Neurophysiol, February 1, 2007; 97(2): 1100 - 1113. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J.-r. Tian, E. Mokuno, and J. L. Demer Vestibulo-Ocular Reflex to Transient Surge Translation: Complex Geometric Response Ablated by Normal Aging J Neurophysiol, April 1, 2006; 95(4): 2042 - 2054. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P Jombik and V Bahyl Short latency disconjugate vestibulo-ocular responses to transient stimuli in the audio frequency range J. Neurol. Neurosurg. Psychiatry, October 1, 2005; 76(10): 1398 - 1402. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Ramat, D. Straumann, and D. S. Zee Interaural Translational VOR: Suppression, Enhancement, and Cognitive Control J Neurophysiol, October 1, 2005; 94(4): 2391 - 2402. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. M. Merfeld, S. Park, C. Gianna-Poulin, F. O. Black, and S. Wood Vestibular Perception and Action Employ Qualitatively Different Mechanisms. I. Frequency Response of VOR and Perceptual Responses During Translation and Tilt J Neurophysiol, July 1, 2005; 94(1): 186 - 198. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W.W.P. Chan and H. L. Galiana Integrator Function in the Oculomotor System Is Dependent on Sensory Context J Neurophysiol, June 1, 2005; 93(6): 3709 - 3717. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P Jombik and V Bahyl Short latency responses in the averaged electro-oculogram elicited by vibrational impulse stimuli applied to the skull: could they reflect vestibulo-ocular reflex function? J. Neurol. Neurosurg. Psychiatry, February 1, 2005; 76(2): 222 - 228. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. Admiraal, N.L.W. Keijsers, and C.C.A.M. Gielen Gaze Affects Pointing Toward Remembered Visual Targets After a Self-Initiated Step J Neurophysiol, October 1, 2004; 92(4): 2380 - 2393. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. R. Barnes and G. D. Paige Anticipatory VOR Suppression Induced by Visual and Nonvisual Stimuli in Humans J Neurophysiol, September 1, 2004; 92(3): 1501 - 1511. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. M. Green and D. E. Angelaki An Integrative Neural Network for Detecting Inertial Motion and Head Orientation J Neurophysiol, August 1, 2004; 92(2): 905 - 925. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. E. Angelaki Eyes on Target: What Neurons Must do for the Vestibuloocular Reflex During Linear Motion J Neurophysiol, July 1, 2004; 92(1): 20 - 35. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Wei, G. C. DeAngelis, and D. E. Angelaki Do Visual Cues Contribute to the Neural Estimate of Viewing Distance Used by the Oculomotor System? J. Neurosci., September 10, 2003; 23(23): 8340 - 8350. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. P. Medendorp, D. B. Tweed, and J. D. Crawford Motion Parallax Is Computed in the Updating of Human Spatial Memory J. Neurosci., September 3, 2003; 23(22): 8135 - 8142. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Ramat and D. S. Zee Ocular Motor Responses to Abrupt Interaural Head Translation in Normal Humans J Neurophysiol, August 1, 2003; 90(2): 887 - 902. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. T. Aw, M. J. Todd, L. A. McGarvie, A. A. Migliaccio, and G. M. Halmagyi Effects of Unilateral Vestibular Deafferentation on the Linear Vestibulo-Ocular Reflex Evoked by Impulsive Eccentric Roll Rotation J Neurophysiol, February 1, 2003; 89(2): 969 - 978. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. P. Medendorp, J.A.M. Van Gisbergen, and C.C.A.M. Gielen Human Gaze Stabilization During Active Head Translations J Neurophysiol, January 1, 2002; 87(1): 295 - 304. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Wiest, J.-R. Tian, R. W. Baloh, B. T. Crane, and J. L. Demer Otolith function in cerebellar ataxia due to mutations in the calcium channel gene CACNA1A Brain, December 1, 2001; 124(12): 2407 - 2416. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Han, J. T. Somers, J. I. Kim, A. N. Kumar, and R. J. Leigh Ocular Responses to Head Rotations During Mirror Viewing J Neurophysiol, November 1, 2001; 86(5): 2323 - 2329. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. P. Medendorp, J.A.M. Van Gisbergen, S. Van Pelt, and C.C.A.M. Gielen Context Compensation in the Vestibuloocular Reflex During Active Head Rotations J Neurophysiol, December 1, 2000; 84(6): 2904 - 2917. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
