|
|
||||||||
Department of Otolaryngology, University of Texas Medical Branch, Galveston, Texas; and Department of Otolaryngology University of Pittsburgh, Pittsburgh, Pennsylvania
Submitted 16 May 2005; accepted in final form 1 September 2005
| ABSTRACT |
|---|
|
|
|---|
| INTRODUCTION |
|---|
|
|
|---|
25 mm from the midline (Huerta and Kaas 1990
Schlag and Schlag-Rey (1987)
redefined the SEF based on the stimulation-evoked eye movements in head-fixed preparations. Since then, the role of the SEF in saccadic eye movement has been extensively studied (Amador et al. 2000
; Bon and Lucchetti 1997
; Chen and Wise 1995
, 1996
; Fujii et al. 1995
; Missal and Heinen 2004
; Russo and Bruce 1993
, 2000
; Schall 1991
; Schiller and Chou 2000
; Schlag and Schlag-Rey 1987
; Stuphorn et al. 2000
; Tehovnik and Lee 1993
). All of these studies were conducted in head-restrained conditions. There were four anecdotal examples provided by Schlag and Schlag-Rey (1987)
, suggesting that SEF stimulation evoked not only eye movements but also head movements.
Several findings emerge in the studies involving SEF stimulation. Like the other oculomotor regions in the cortex, the evoked saccades in the SEF were directed toward the side contralateral to the stimulated hemisphere. In the large-amplitude sites, the stimulation-evoked saccades often converged toward a region located on the contralateral side (Russo and Bruce 1993
; Schlag and Schlag-Rey 1987
; Tehovnik and Lee 1993
). When the horizontal initial eye positions (IEPh) were deviated toward the ends of the converging saccades, the probability of stimulation-evoked saccades decreased and the threshold current (i.e., the stimulation current that evoked eye movements 50% of the time) increased up to 100 µA or higher. In some cases, eye movements were not elicited even at the highest current levels tested (Schlag and Schlag-Rey 1987
; Tehovnik and Lee 1993
; Tehovnik and Sommer 1997
). When saccades did occur, their onset latencies increased. It was also reported that, when the eyes were deviated in the unresponsive initial eye positions (IEPs) at target onset, stimulating the SEF prevented the saccadic execution (Tehovnik and Lee 1993
). This unresponsive oculomotor space often covered a large contralateral field or occasionally up to the entire contralateral hemisphere. Whether head movements could be evoked within this relatively unresponsive oculomotor region remains unknown.
Recent developments in movement control techniques have made it possible to adequately assess the stimulation-evoked movements in head-unrestrained monkeys (Corneil et al. 2002a
,b
; Freedman and Sparks 1997
; Phillips et al. 1995
; Sparks et al. 2001
). A recent microstimulation study investigated the SEF in head-unrestrained monkeys (MartinezTrujillo et al. 2003
). This study observed that SEF stimulation evoked coordinated movements of the eyes and head. This study also noted no sign of site-specific clustering for particular movements, e.g., segregation of eye- and head-movement sites. Instead the SEF stimulation evoked both eye and head movements that were kinematically similar to the visually guided eye-head movements.
If the SEF plays a role in controlling movements of the eyes and head, the following issues can be raised. First, are the stimulation-evoked head movements in the SEF necessarily accompanied by gaze shifts? It has been shown that subthreshold stimulation in the superior colliculus, which is interconnected with the SEF, can evoke slow head movements in the absence of gaze (monkey: Corneil et al. 2002a
,b
; cat: Galiana and Guitton 1992
; Pelisson et al. 2001
). It is not known whether, under similar or different circumstances, the SEF stimulation evokes similar types of head movements. Second, do the stimulation-evoked head movements in the SEF follow the rules established under the visually guided conditions, i.e., is the relative contribution of the eyes and head to the gaze shift affected by the initial positions of the eyes in the orbits (Delreux et al. 1991
; Freedman and Sparks 1997
; Fuller 1992
; Gandhi and Sparks 2001
; Goossens and Van Opstal 1997; Guitton et al. 1990
; Sparks et al. 2001
; Tomlinson and Bahra 1986
; Volle and Guitton 1993
)? Third, does SEF stimulation affect a process that is unique to head-movement control following gaze shifts, i.e., centering the eyes in the orbits (Sparks et al. 2001
)? Previous research has shown that a short-duration of microstimulation truncates movements prematurely (for review, see Graziano et al. 2002
). When the stimulation duration is extended to permit movement completion, complex movement sequences rather than muscle twitches or movement segments could be observed. Previous SEF stimulation studies in head-unrestrained conditions had used relatively short stimulation trains (160200 ms) (Martinez-Trujillo et al. 2003
; Schlag and Schlag-Rey 1987
); the stimulation was terminated before the head movement was completed. Whether extending the stimulation in the SEF affects the postgaze-shift head movements, which in turn facilitate centering the eyes in the orbits, is not known.
The present study was designed to address our questions by employing stimulation of the SEF in head-unrestrained monkeys. Stimulation duration and the initial eye positions were varied systematically. The present data showed that stimulation in the SEF evoked head movements in the absence of gaze on some trials, particularly when the eyes were deviated in the contralateral direction at stimulation onset. Increases in stimulation duration resulted in increases in the duration of the head movements during the postgaze-shift periods. This portion of the head movements contributed significantly to centering the eyes in the orbits. The present results suggest the SEF plays a role in head-movement control independent of gaze.
| METHODS |
|---|
|
|
|---|
Two juvenile rhesus monkeys (Macaca mulatta, 57 kg) served as subjects. In each monkey, a scleral coil was implanted over one eye to monitor eye movements (Fuchs and Robinson 1966
; Judge et al. 1980
). A different search coil, serving to monitor head movements, was cemented above the parietal bone near the midline. A stainless steel head-post was implanted over the occipital bone to restrict the animal's head movements during part of the experiments. A separate stainless steel post was implanted on the anterior ridge of the frontal bone; this post served to secure a lightweight microlaser (Edmund Scientific, No. M52263) fixture and juice-delivering tubing. A rectangular recording chamber (36 x 28 x 6 mm) was implanted and cemented over the left side of the frontal bone (17 mm A-P and 12 mm M-L). During recording, a cylindrical adaptor (18 mm diam) was used to fix the microdrive and the micropositioner to the recording chamber at the known coordinates to facilitate access to the SEF.
The monkeys were seated in a primate chair for daily training and experiments. The monkeys' heads were placed in a position near the center of the magnetic field. On head-restrained conditions, the head-post was clamped to a fixture that was secured to a rigid bar attached to the frame of the primate chair. On head-unrestrained conditions, the head post and the clamping bar were removed. A primate chair designed for conducting the head-unrestrained monkey experiment was used. The typical, horizontal neck plate that prevented the animals from moving their heads vertically was removed. Instead, the monkeys' collar (Primate Products) was secured between two pieces of paralleled Plexiglas plates, one against the animals' chests and the other against the animals' backs. When the monkeys' heads were free to move, the fore-aft displacements of their upper torsos and shoulders were limited to <2 cm between the two plates. Given these constraints, the head rotations were likely accounted for by neck rotations in the cervical segments of the spine. Contribution of the rest of the body to head rotations was minimal. The range of the head rotations was approximately ±50° along the horizontal meridian and ±30° along the vertical meridian. All surgical and experimental procedures conformed to the guidelines for the Care and Use of Animals of National Institutes of Health and the Institutional Animal Care and Use Committee.
Data acquisition
A 42-in cubic coil and a phase-angle detection system (CNC Engineering) were used to measure the horizontal and vertical position signals of the gaze (eye re space) and head sampled at 500 Hz. This system, in the horizontal dimension, was linear within 2% error over the entire range. Signals of the gaze coil were calibrated in the head-fixed condition by rewarding the monkeys for aligning their line of sight with the visual targets. In head-unrestrained conditions, the signal of the eye coil would reflect the line of sight with respect to the space (i.e., gaze) not with respect to the head. Signals of the head coil were calibrated in a head-fixed condition in which the overhead microlaser was turned on and the microlaser beam (with the head) was aligned with the visual targets of known angles. The position of the eyes (re head) was computed off-line by mathematically converting the horizontal and vertical coil signals to unit vectors, and the vectors were rotated with respect to the head in Fick coordinate. All of the displacement measurement was carried out following vector rotations by assuming zero torsion (Haslwanter 1995
; Straumann et al. 1991
).
A hydraulic microdrive (Kopf Instruments) was used to advance the epoxylite-insulated tungsten electrodes (Frederic Haer) through the dura surface, which was prepunctured by a sharpened needle (details see Chen et al. 2001
). Neural signals were band-pass (500 Hz to 5 kHz) filtered using a differential amplifier (Bak Electronics). Microstimulation was carried out using a stimulator (S88, Grass Instruments) and an optical isolation unit (PSIU6, Grass Instruments). The stimulation pulses were discriminated using a BAK window discriminator. The stimulation trains consisted of 0.2-ms, monopolar, cathodal pulses. Stimulation duration ranged from 300 to 600 ms. Because there was no previous report regarding the current threshold of the stimulation-evoked head movements in the SEF, we typically explored with a two to three times suprathreshold current identified for evoking gaze shifts. Typical current intensity and stimulation frequency were 100 µA (range: 80150 µA) and 200 Hz (range: 100200 Hz), respectively. We found that the current of 100 µA was, in general, effective in evoking eye-head combined gaze shifts. It was difficult to monitor the actual current delivered through the high-impedance electrodes; all of the current intensity reported in the present report was taken from the face value of the stimulator.
Data were acquired using a Pentium microcomputer running an in-house data acquisition software originally developed by P. Glimcher and D. Sparks. This data-acquisition program allowed a laboratory computer to control the location and duration of visual targets and to monitor horizontal and vertical positions of gaze and head. In addition, the software allowed for controlling the delivery of the stimulation trains and the juice reward.
Behavioral paradigms and microstimulation
Visual targets were presented on a tangent screen with a rectangular array of tri-state (red, green, yellow) light-emitting diodes (LEDs), which consisted of 41 rows of 49 lights equally spaced (by 2 in) in either horizontal or vertical dimension. Throughout the experiments, the LED board was placed at a distance of 72 cm (28.5 in) from the monkeys. The spacing between the adjacent horizontal or vertical LEDs near the center of the board was
2° with respect to the monkeys.
The monkeys were trained in a visually guided gaze shift task that permitted independent control of the gaze and head. Figure 1A illustrates the onset and offset sequence of the targets in the task. The task began with the monkeys sitting straight ahead, aligning a head-mounted microlaser beam with a visual (red) target via visual feedback. Later, a second (green) target was illuminated, and the monkeys deviated their eyes to the target while maintaining a stable head posture. The green target was usually displayed in 27° steps (i.e.,
15 LED in spacing) horizontally or vertically with respect to the horizontal or vertical meridian of the initial red target. Some 600800 ms after the onset of the green target, all targets and the microlaser were extinguished for 400500 ms. The same red and green targets and the microlaser were re-illuminated (500700 ms) and then re-extinguished. Some 400600 ms later, a third (yellow) target was illuminated at a randomly chosen location, and the monkeys made a gaze shift to acquire the target. A drop of juice reward was delivered in contingency on the successful execution of the gaze shift. To achieve eye-head postural constraints, gaze and head "windows," typically ranging 5 and 10° in radius, respectively, were set up in a close-loop real-time control. Reward was contingent on the monkeys keeping their gaze and head positions within these windows. If either gaze or head stepped outside of the windows during the designated period of postural constraints, the trial was aborted.
|
Electrical microstimulation trials (Fig. 1B), interleaved with control trials, were carried out in
50% of the trials for each block. The stimulation trials were similar to the control trials except that the electrical stimulation was always delivered 200 ms after the first extinction of both the red and green targets. From that point on, the red and green visual targets and the microlaser remained extinguished. Approximately 8001,400 ms later, a visual (yellow) target was illuminated, just like that in the control trials, and the monkeys were rewarded for making a gaze shift to the target. A major difference between the stimulation and the control trials was that the former imposed postural constraints up to the point before stimulation was delivered, whereas the latter imposed postural constraints throughout the task.
The control trials were designed to mimic the stimulation trials. Both trial types imposed that the monkeys maintained stable eye and head postures in darkness after the extinction of the red and green targets and the microlaser. Two critical features in these designs are worth of pointing out. The red and green targets were re-illuminated 500700 ms after the beginning of the gap period (in the control trials), and the stimulation was always applied 200 ms after the beginning of the gap period (in the stimulation trials). Therefore at the time of stimulation, there was no way for the monkeys to know whether the trial was a stimulation trial or a control trial. The monkey could only anticipate the re-appearance of the red and green targets at locations corresponding to their current eye and head positions. In other words, re-illuminating the visual targets in the control trials ensured that the animals were motivated to maintain their current eye and head positions rather than anticipating and executing volitional or anticipatory movements.
In some stimulation sites, we carried out the "nontask mode" stimulation, in which the stimulation train was delivered during the inter-trial interval (duration: 5001,500 ms; see RESULTS). Like the "task mode" stimulation, there was no visual target and no postural constraints prior to, during, or after the delivery of the stimulation. However, a visual (yellow) target was illuminated after a similar time course as the task-mode stimulation, and the monkeys were rewarded for making a gaze shift to the target. There were two important aspects of this trial type. First, the stimulation was delivered without warning, i.e., there were no visual or auditory stimuli that cued the animals as to when the trial started. Even if the animals wanted to, they could not time the upcoming stimulation. Second, there was no postural restriction for the positions of the eyes or head. The animals were presumably as relaxed as they could be in any natural upright postures. It was unlikely for the animals to generate volitional movements on a consistent basis within tens of milliseconds after the stimulation.
Due to the fact that it was impossible to know whether a given electrode penetration was made parallel, orthogonal, or oblique with respect to the cortical surface, we considered the stimulation depth separated by 500 µm as different sites. Concerning the fact that SEF was located on top of the dorsomedial frontal cortex and was relatively shallow, in general, a successful electrode penetration yielded one stimulation site with usable data. Occasionally, two sites with distinctive characteristics were obtained in the same penetration. The evoked movements in different sites were analyzed separately. In addition, care was taken to avoid stimulating the white matter beneath the cortex. The electrodes were always advanced sufficiently deep, and the border between gray and white matter was identified based on the diminishing unit discharge in the background. Once the electrode was confirmed to have reached the white matter, the electrodes were then slowly withdrawn. The stimulation was carried out
100 µm away from the border of the white and gray matter. Data obtained in the white matter, if any, were excluded from the analyses.
Data analyses
Data analyses were performed using an in-house program on a Windows platform. Movement onsets and offsets were determined based on a set of velocity criteria (gaze: 80°/s for the horizontal onset and 60°/s for the vertical offset; head: 6°/s for both horizontal and vertical onset and offset). Movements were displayed 100 ms prior to and 800 ms after the stimulation onset, and measurements were taken strictly based on velocity criteria. Any movement detected prior to stimulation onset was removed from further analysis. Due to the inherent noise in the low-velocity head-movement data, the head-movement velocity was filtered separately from the gaze data. A digital Taylor-series filter and a sliding probabilistic threshold-window were implemented. The latter, probabilistic threshold-filter computation was designed to circumvent the situation in which the head velocity fluctuated around the velocity threshold (Chen and Walton 2004
). For the head-movement onset velocity cutoff, a 36/50 ratio (i.e., 72 vs. 100 ms) was used. For head-movement offset velocity cutoff, a 11/16 ratio (i.e., 22 vs. 32 ms) was used. In our experience, there were only a few cases, if any, where the onset of the head movements was ambiguous. These head movements often had an inconsistent velocity profile near or below the detection threshold, such that no human intervention would make much difference. These trials were abandoned.
Staircase gaze shifts were occasionally encountered in the SEF stimulation data. These gaze trials could be accompanied by multiple head movements that were not within the scope of the present study. These trials were excluded from the analyses. Statistical analyses were performed using Statistica (Statsoft). Throughout this paper, the averages of the data are presented as means ± SD unless otherwise indicated.
At the end of experiments, the monkeys were killed with an overdose of pentobarbital, and the brains were removed for histological examination. Stainless-steel pins were inserted through the known coordinates and left in the brain during perfusion for the facilitation of coordinate reconstruction.
| RESULTS |
|---|
|
|
|---|
10° when the eyes deviated in the direction ipsilateral to the stimulated side) were selected for conducting the stimulation experiment as large-amplitude gaze shifts would recruit significant head movements (Guitton et al. 1990
Figure 2 illustrates the horizontal position (A) and velocity traces (B) of the stimulation-evoked [gaze, eye (re head), and head] movements in a given SEF site. Because the head velocities were typically lower than those of the gaze and eye movements, the peaks of gaze- and eye-velocity traces were truncated to facilitate the display of head-velocity profiles (Fig. 2, BD). As shown in the velocity traces in Fig. 2B, the gaze-, eye-, and head-velocity traces initially overlapped one another. About 70 ms after the onset of stimulation, the head velocity increased to
20°/s. At the same time, the eye velocity increased in synchrony with the head by approximately the same magnitude but opposite in direction (i.e., VOR gain equals one). Some 100 ms later, the gaze shift was initiated; head velocity increased to
4050°/s. Head velocity peaked after the offset of gaze shift and eventually declined to
0. As in the examples shown in Fig. 2, the entire movement was accomplished within 300400 ms. The gaze shifts, which occurred with the head movements, were as brief as 50100 ms. Furthermore, the onset latency of the head movements could vary relative to either the onset of stimulation or the onset of gaze shifts. Head onset preceded gaze onset by 60100 ms as was the case in Fig. 2, B and D. By contrast, head onset lagged gaze onset by
35 ms in Fig. 2C.
|
We often observed eye and head, head alone, and eye alone trials in the same site. There was no indication of any site-specific clustering for the movements evoked by electrical stimulation. However, depending on the context of movement execution (e.g., stimulation duration and eye positions in the orbits), the characteristics of the stimulation-evoked movements would vary. These points will be elaborated in later sections.
Kinematics of the stimulation-evoked head movements
The trials in which the initial head position was centered with respect to the body (see METHODS) and the head displacements were
2° (n = 1,163) were selected for the kinematics analysis. Figure 3A plots the relationship between the horizontal and vertical components of these head movements, in which two features can be noted. First, the overwhelming majority (99.1%, 1,153/1,163; M1 = 1,074/1,084; M2 = 79/79) of these head movements were directed away from (i.e., contralateral with respect to) the stimulated side. The rest of the head movements had very small displacements (2.6 ± 1.2°) in the ipsilateral direction.
|
Figure 3B plots the main sequence characteristics, i.e., peak velocity versus displacement, of these head movements. The slope of the linear regression of the data were 3.86 (Pearson Correlation r = 0.90; P < 0.01), which closely resembled the main sequence of the visually guided head movements. When the data were separated for IEPh conditions, the slope in the IEPo condition (slope = 3.5, r = 0.89) was slightly lower than that in IEPi (slope = 4.2, r = 0.95) or IEPc (slope = 4.7, r = 0.89) conditions (ANCOVA for homogeneity of slope model, F = 7.8, P < 0.01).
The horizontal peak head velocity (Fig. 3B) was dwarfed when compared with the horizontal peak eye (re head) velocity (Fig. 3D). In general, the horizontal peak head velocity was highly limited, e.g.
1/6 of the magnitude of the horizontal peak eye velocity for a 20° horizontal eye displacement. These data were consistent with those of visually guided head movements (Corneil et al. 2002a
,b
; Freedman and Sparks 1997
; Guitton et al. 1990
; Martinez-Trujillo et al. 2003
; Phillips et al. 1995
).
Effects of stimulation duration
To gain insights into the detailed characteristics and timing of the stimulation-evoked movements, Fig. 4 plots individual traces of the horizontal head and gaze velocities for a given SEF site. The top panels illustrate the example head-velocity traces evoked by 300-ms stimulation, and the bottom panels illustrate the traces evoked by 600-ms stimulation. The plots are grouped by IEPh conditions. Five traces (2 of high peak velocities, 2 of low peak velocities, and 1 of medium peak velocity) were selected to represent the typical head- and gaze-velocity profiles for a given stimulation condition. Several features can be noted in the plots. First, similar to those observed in the visually guided eye-head combined movements, the stimulation-evoked head velocity typically rose to peak near or following the end of gaze shifts (i.e., when the gaze velocity declined to near baseline). The peak velocities of the head movements varied from
20 to
120°/s. However, for a given IEPh condition, there was no consistent trend that could predict the peak head velocity on a particular trial.
|
2030°/s within 100200 ms after gaze offset. The head velocities were maintained at this low velocity until near or beyond the end of the stimulation train. In many cases, the head movements continued as long as the stimulation continued. This observation was consistent across all IEPh conditions. Third, head onset varied significantly with respect to stimulation onset. Also, head onset varied significantly with respect to gaze onset. The variability of head onset persisted from trial to trial in all IEPh conditions. A close examination of this figure suggests that the stimulation often produced a low-velocity phase of head movement that remained below threshold for up to several tens of milliseconds. Unfortunately, this early phase of the movement was sometimes of such a low velocity that it was impossible to reliably detect. Nonetheless, it is clear that the stimulation did not consistently produce a robust head movement beginning at a particular latency.
Fourth, significant head movements were sometimes evoked in the absence of gaze shifts. These head-alone movements were fairly common in the IEPc condition (93%, 1,000/1,074). Head-alone movements were also sometimes observed in the IEPo condition (114/212). In sharp contrast, the majority (97%, 289/297) of the trials obtained in the IEPi condition contained eye-and-head movements. In general, the peak head velocities of the head-alone trials were comparable to those of the eye-and-head trials. The biased occurrence of the head-alone trials in the IEPc condition was consistent with the notion that the majority of SEF sites contained unresponsive IEP regions in the direction contralateral to the stimulated sides (Schlag and Schlag-Rey 1987
; Tehovnik and Lee 1993
). Also, the existence of the head-alone trials suggests that gaze shifts were not a prerequisite for the stimulation-evoked head movements.
Onset and offset latencies of head movements
To quantify the effects of stimulation duration on the onset and offset latencies of head movements, the data obtained in two SEF sites were selected for plots in Fig. 5. It can be noted that the distributions of the head onset latencies for 300-ms [117 ± 44 ms (n = 50) and 120 ± 52 ms (n = 48) for sites A and B, respectively] and 600-ms [107 ± 39 ms (n = 30) and 134 ± 54 ms (n = 41) for sites A and B, respectively] trials overlapped each other. No significant differences were found between the two trial types (F = 1.5, P = 0.22 for Fig. 5A1; F = 1.1 P = 0.29 for Fig. 5B1).
|
Figure 6 quantifies the latencies of head onset as a function of IEPh across all of the stimulated sites in the SEF. When head movements with onset latencies within 50300 ms were selected for this analysis, there was a significant correlation between the head onset latencies and the IEPh [slope = 1.13 (M1 = 1.11; M2 = 1.14); r = 0.32, F = 178, P < 0.01; Fig. 6]. When the IEPh was deviated in the direction opposite of the movements (i.e., IEPi; negative IEPh values), the latency of head onset was increased as compared with the other conditions. Post hoc analyses revealed that the head onset latency in the IEPi condition (mean latency: 155 ± 63 ms, n = 297) was significantly longer than those in IEPo (mean latency: 123 ± 68 ms, n = 326) and IEPc (mean latency: 111 ± 46 ms, n = 493) conditions (ANOVA, Scheffe test, P < 0.01 for both). There was no significant difference in the latencies of head onset between IEPo and IEPc conditions (Scheffe test, P = 0.31).
|
|
As the stimulation-evoked gaze shifts often drive the eyes to the eccentric positions in which the visual and oculomotor performances were limited, one may wonder whether the eye positions in the orbits are correlated with the displacements of the postgaze-shift head movements. Figure 8 illustrates the examples of the effect of eye positions at gaze offset on the displacement of the postgaze-shift head movement. The displacement of the postgaze-shift head movement, measured as the head displacement between gaze offset and head offset during which the gaze was stabilized (Fig. 8A), will be referred to as "head contribution to eye counter-roll (HcEc) during the postgaze-shift period" in the remainder of the text. Several observations are described in the following text.
|
Figure 9 quantifies this effect across all SEF sites tested. The analysis was carried out on the eye-and-head trials among those analyzed in Fig. 7. For the sake of comparison, the trials in which gaze shifts were most consistently observed across three different stimulation durations (i.e., IEPi and IEPo conditions) were pooled together for analysis. Overall, the horizontal head contribution to eye counter-roll was strongly correlated with the horizontal eye position at gaze offset (Fig. 9A; r = 0.59, P < 0.01, n = 560). Note that in Fig. 9A, when EPh was
0° (i.e., ipsilateral to the stimulated side), the head contribution to eye counter-roll was negligible (0.5 ± 0.8°, n = 27). In other words, when the eyes were already near the center of the orbits, stimulation failed to drive the head to continue moving. But as the horizontal eye positions were deviated further contralaterally, the head contribution to eye counter-roll was increased systematically as a function of the horizontal eye positions at gaze offset.
|
0°, 300-ms stimulation consistently evoked relatively less head contribution to eye counter-roll, compared with either 500- or 600-ms stimulation trials (Scheffe test, P < 0.02). That is, extended stimulation prolonged the head movements and thus improved the head contribution to eye counter-roll. As pointed out earlier (Fig. 3), the stimulation-evoked head movements in the SEF were primarily horizontal. The vertical head contribution to eye counter-roll in these trials was negligible (average: 0.2 ± 0.7°) regardless of stimulation durations (r = 0.01, P > 0.80; Fig. 9B).
Head-alone trials
In these trials, the correlation between horizontal head contribution to eye counter-roll and eye position at head onset was relatively weak but remained significant. Figure 10A illustrates the example of a positive correlation between the horizontal head displacement and the horizontal eye positions (r = 0.60, P < 0.01, n = 53 for site B; r = 0.58, P = 0.001, n = 28 for site C) in two SEF sites. The stimulation duration had a significant interactive effect in these trials (F = 17.5, P < 0.01 in Fig. 10B; F = 8.9, P < 0.01 in Fig. 10C), indicating that extended stimulation indeed improved the horizontal head contribution to centering the eyes in the orbits.
|
5°, there was no significant difference between these trials (Scheffe test, P > 0.23).
|
Several unique differences could be noted in the comparison between the eye-and-head trials (Fig. 9A) and the head-alone trials (Fig. 11A). First, the former contained a negligible horizontal head contribution to eye counter-roll when EPh was
0°. The latter contained a small but consistent horizontal head displacement (4.1 ± 2.2°, n = 76) even when EPh was centered in the orbit (range: 0:5°) or biased ipsilaterally (range: 5:0°; indicated by a downward arrow in Fig. 11A). The difference between the two trial types was highly significant (t-test, P < 0.01). This unique feature was reminiscent of the stimulation-evoked head-alone movements in the superior colliculus and had not been reported in the SEF in the past (Corneil et al. 2002a
,b
; Pelisson et al. 2001
). Second, the head-alone trials (Fig. 11A) did exhibit a general tendency for the horizontal head displacement to increase as a function of the horizontal eye positions. However, detailed inspection indicated that this effect was not significant between when the eyes were deviated 2025° in the orbits and when the eyes were deviated 2530° in the orbits (F = 0.02, P > 0.86 for 600-ms trials; F = 0.03, P > 0.87 for 500-ms trials; F = 3.6, P > 0.06 for 300-ms trials; Fig. 11A). It appears that the differences occurred depending on whether or not gaze shifts took place.
Head contribution to gaze shift
Figure 12 illustrates the effect of IEPh on the relationship between head contribution to gaze shift and gaze amplitude. The horizontal head contribution to gaze shift (HcGh; defined as the horizontal head displacement between gaze onset and gaze offset; Fig. 12A) was positively correlated with horizontal gaze (Gh) amplitudes (Fig. 12B). This relationship shifted when the IEPh was systematically varied. The vertical head contribution to gaze shift (HcGv, defined as the vertical head displacement between gaze onset and gaze offset) was negligible (0.32 ± 0.77°; slope = 0.000.01; Fig. 12C). These results were reminiscent of those observed in the visually guided gaze shifts (Guitton et al. 1990
; Phillips et al. 1995
; Sparks et al. 2001
; Tomlinson and Bahra 1986
).
|
|
1° were considered, a trend becomes clear and can be best appreciated in the box plots. In IEPi condition, the middle 50% of the data were centered about zero (range: from 14 to 18 ms, n = 165). That is, as long as the head and gaze movements took place within 20 ms of each other, significant head contribution would be recruited. However, in IEPo and IEPc conditions, the middle 50% of the data ranged between 139 and 44 ms (n = 56) and between 188 to 130 ms (n = 15), respectively. In other words, if the head was already moving when gaze shifts began, the head often made a significant contribution to gaze shifts.
We further contrasted the timing difference between the head and gaze onsets of the stimulation trials with that of the visually guided gaze shift trials. The latter trials took place following the illumination of the yellow target in the task (see Fig. 1 and METHODS). This timing difference in the visually guided gaze shifts ranged from 204 to 196 ms with an average of 5.1 ± 43.4 ms (n = 1370; 5.2 ± 39.6 ms for M1, n = 638; 4.9 ± 46.6 ms for M2, n = 732). Similar to the finding illustrated in Fig. 13, when the horizontal head contributions to gaze shifts of
1° were considered, the values leaned toward the negative end, i.e., the difference between head onset and gaze onset became 20.8 ± 29.2 ms (19.1 ± 24.9 ms for M1; 22.2 ± 32.4 ms for M2). In essence, consistent with previous studies, the significant head contributions to gaze shifts in these visually guided trials came primarily from those in which head onset led gaze onset (Galiana and Guitton 1992
; Gandhi and Sparks 2001
; Goossens and Van Opstal 1997
; Guitton et al. 1990
).
Eye-alone trials
As pointed out earlier, SEF stimulation sometimes evoked gaze shifts in the absence of head movements. The head movements in these eye-alone trials (n = 543) were simply nonexistent or below the velocity criteria (see METHODS). The majority (81%; n = 438; M1 = 419/522; M2 = 19/21) of the eye-alone trials were obtained in the IEPi condition, whereas the minority (19%; n = 104; M1 = 102/522; M2 = 2/21) was obtained in the IEPo condition. In the IEPi condition (average IEPh: 23.0 ± 2.7°), the average horizontal gaze displacement was 22.5 ± 7.4°, and the average EPh at horizontal gaze offset was 0.2 ± 6.6°. In the IEPo condition (average IEPh: 0.2 ± 1.5°), the average horizontal gaze displacement was 7.7 ± 5.0°, and the average horizontal eye position (EPh) at gaze offset was 7.2 ± 4.6°. Only one eye-alone trial was found in the IEPc condition; this trial had a horizontal gaze displacement of 3.4°. In essence, the movement metrics exhibited in the eye-alone trials had two main features: the horizontal gaze displacements remained comfortably within the oculomotor range and the eye deviations at the end of the gaze shifts remained close to the center of the orbits. In such a case, head movements were not recruited as shown in visually guided gaze shifts (Freedman and Sparks 1997
; Guitton et al. 1990
; Sparks et al. 2001
).
Nontask mode stimulation
One may wonder whether the visual stimuli or the postural control in the task had biased the outcome of SEF stimulation. Specifically, one may wonder whether the observed early head movements had been voluntary movements initiated by the animals. To rule out these possibilities, we conducted the nontask mode stimulation in 18 SEF sites (16 in M1 and 2 in M2) during the intertrial interval in the dark (see METHODS).
Figure 14A illustrates the stimulation-evoked movement traces under this trial type in a SEF site. Based on the comparison between the traces of Fig. 14A and those of Fig. 4, one can note that these trials were alike in many aspects. Head onset could lag or lead gaze onset (Fig. 14A, 300-ms stimulation). The head-alone trial could be evoked occasionally (Fig. 14A, middle traces in 600-ms stimulation). Also, the early head trials were prevalent. These trial types were present in both monkeys.
|
; 600-ms trials:
) were comparable to that illustrated in Fig. 9. The stimulation duration had a significant interactive effect over the head contribution to eye counter-roll (ANCOVA, separate slope model; F = 42.8, P < 0.001).
We further compared the stimulation-evoked head movements to the visually guided head movements. Figure 14B illustrates the average (
) and the range (
) of head contribution to eye counter-roll for the visually guided gaze shift trials (n = 1,370; n = 638 for M1 and n = 732 for M2). Two major points can be noted. First, the data points of both 300- and 600-ms stimulation trials fell within the range of the visually guided gaze shift trials. Second, consistent with the fact that 300-ms stimulation prematurely truncated the head movements, the 300-ms stimulation trials exhibited smaller head contribution to eye counter-roll as compared with the average of the visually guided gaze shift trials. On the other hand, 600-ms stimulation exceeded the completion of typical head movements, and thus these trials exhibited higher than average head contribution to eye counter-roll.
SEF map
Figure 15 summarizes the distributions of eye-alone, head-alone, and eye-and-head trials in the SEF of one monkey in which the SEF was thoroughly mapped. Only movements of
2° were included in this analysis. It can be noted that the eye-alone trials were relatively evenly distributed (Fig. 15B). The center of gravity for these sites was
7 mm rostral from the caudal end of the arcuate sulcus and
4 mm from the midline. This SEF map appears consistent with previous studies that suggest that the saccadic sites are located rostrally whereas the smooth pursuit sites are located caudally (Chen and Wise 1995
; Fujii et al. 2002
; Fukushima et al. 2004
; Schall et al. 1993
).
|
| DISCUSSION |
|---|
|
|
|---|