| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Optometry and Neuroscience, University of Manchester Institute of Science and Technology, Manchester M60 1QD, United Kingdom
Submitted 19 December 2002; accepted in final form 20 June 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
; Ohashi and Barnes 1996), and hence bring the target image close to the fovea, reduce blur, and aid perception of fast moving objects. In humans, with regularly timed, repeated presentation of target motion stimuli, smooth pursuit precedes target onset by several hundred milliseconds (Barnes and Asselman 1991; Kao and Morrow 1994). Anticipatory smooth pursuit can be elicited having viewed a moving target while the eyes are fixated on a central target (Barnes et al. 1997) and can be scaled by symbolic cues regarding upcoming target velocity (Jarrett and Barnes 2001). These findings indicate that anticipatory smooth pursuit is not simply a direct consequence of an efference copy of the eye movement. The initiation of smooth pursuit before target onset can also be generated by context-dependent cues and has been interpreted to be the result of sudden activation of an on-line gain-controlling element (Tanaka and Fukushima 1997; Tanaka and Lisberger 2000).
The influence of the above factors has led to ocular pursuit being modeled as a system that receives both retinal and extraretinal inputs (Barnes and Asselman 1991; Barnes et al. 1995; Krauzlis and Lisberger 1994; Krauzlis and Miles 1996; Leigh and Zee 1991; Robinson et al. 1986). The retinal input is derived from direct feedback of visual motion signals such as image velocity and acceleration (Krauzlis and Lisberger 1994), whereas the extraretinal input is influenced by factors such as expectation, attention, memory, and the sensory consequences of eye motion. In combination, these inputs to the visuomotor drive enable smooth pursuit gain close to unity during steady-state pursuit of a moving target, eliminating the lag between the eye and target that would occur if the response were driven by visual feedback alone.
When the image of a moving target is stabilized on the retina, smooth pursuit continues, but with a reduced gain because of the lack of input from visual feedback (Morris and Lisberger 1987; Pola and Wyatt 1997). Similarly, when a moving target suddenly disappears from view, smooth pursuit continues at a reduced gain only if subjects expect the target will reappear (Becker and Fuchs 1985) or if they direct attention to "pushing" the imagined target (Pola and Wyatt 1997). Without the influence of expectation or attention, eye velocity decays to zero in roughly an exponential manner (Mitrani and Dimitrov 1978). When the target is expected to reappear, smooth pursuit in the complete absence of visual feedback can be maintained for relatively long durations. Becker and Fuchs (1985) found that on average eye velocity decayed to a reduced level (i.e., residual velocity) within 450 ms, and was then maintained for as long as 4,000 ms when subjects were instructed to continue pursuit of a previously visible moving stimulus so that they were on target when it reappeared. Although unsubstantiated, the investigators commented that in some of their preliminary work, the eye tended to reaccelerate before target reappearance when the nonvisible target duration [interstimulus interval (ISI)] was constant and hence predictable. With the randomized target velocities (i.e., nonpredictable) there was an increase in eye velocity back toward 100% only after target reappearance. The authors speculated that expectancy regarding target reappearance influenced the gain of the pursuit mechanism, causing the increase in eye velocity toward the end of the transient. There was no attempt to model the structure of the pursuit mechanism, such that it would provide a realistic simulation of the temporal development of eye velocity.
If appropriately timed, an increase in eye velocity before target reappearance could compensate for the decay that follows the loss of visual feedback during the transient disappearance of a moving target. However, given the potential range of ISIs (400 to 4,000 ms; Becker and Fuchs 1985), it follows that the mechanism controlling any increase in eye velocity should ideally be linked to the moment of target reappearance. Perceptual experiments have indicated that information on the time of reappearance can be inferred from target motion characteristics before a nonvisible part of a trajectory and is used in making anticipatory judgments (Lyon and Waag 1995; Peterken et al. 1990). It remains to be empirically verified whether subjects can use the predictable timing information, provided when the target velocity and ISI are constant, to control eye velocity before target reappearance.
The general aim of the present study was to examine subjects' ability to extrapolate pursuit over a transient period of nonvisible target motion. To this end, we examined the influence of stimulus predictability, target velocity, and ISI. We then constructed a model of pursuit, bounded by previously reported experimental data, which simulated the temporal development of eye velocity during the transient disappearance of a visual target.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
A total of 9 subjects participated (mean age 31.8 yr; SD 10.3 yr), 4 of whom were completely naive of oculomotor experiments. All subjects had normal or corrected-to-normal vision, were healthy, and had no relevant medical or psychiatric history. The experiments were conducted according to a protocol approved by UMIST local ethics committee in conformity with the tenets of the Declaration of Helsinki. All subjects participated with informed consent.
Apparatus
The experiment was conducted in a purpose-built dark room. Subjects were seated centrally, in front of a flat white screen (1.5 x 1.5 m) at a viewing distance of 1.7 m. The head was supported on an adjustable chin rest and fixed by clamps to the sides. The visual target consisted of a ring of 12 light-emitting diodes (LEDs) that were optically reduced to form a ring of dots subtending 1.2° on the screen. When projected on the screen the LEDs had a luminance of 0.5 cd/m2. Subjects reported no difficulty seeing the target. Although differing from previous research that has used a single dot (Barnes and Asselman 1991) or cross (Barnes et al. 1987), it has been shown that multiple-dot stimuli are sufficient to drive smooth pursuit (Heinen and Watamaniuk 1998).
The horizontal motion of the target was controlled by reflection from a mirror galvanometer. Toggling the illumination of the LEDs controlled the target visibility. Eye movements were recorded by an infrared limbus tracking technique (Skalar IRIS) that was mounted to an adjustable head strap. A bite bar mounted to the head strap was used to eliminate movement of the device once it had been calibrated to an individual subject. The output from the device was sampled at intervals of 4 ms and low-pass filtered at 80 Hz (4-pole Bessel) before storage on disc and off-line analysis. Before each trial a calibration was performed in which subjects pursued a sinusoidal horizontal oscillation at a frequency of 0.4 Hz with amplitude of ±20°. At the end of the calibration the target remained stationary at the center position for 2,500 ms, during which subjects maintained fixation. The corresponding eye position was centered to the known midpoint of the target displacement to reduce the influence of any recording error at the extremities of eye movement. The resolution with which eye position was recorded was about 10- to 20-min arc. A calibration was deemed successful when the linearity between the eye and target signal was better than 5%.
Procedures
Subjects performed 4 control trials and 6 experimental trials, each consisting of 18 separate presentations. Subjects were required to track a horizontally moving target as accurately as possible throughout each presentation. The start of a presentation was signaled by an auditory warning cue of 80-ms duration. Simultaneously, the target was illuminated and remained stationary at a position of –20° to the left of the screen center for a duration of 960 ms. In 2 control trials (type I) the target was extinguished for 420 ms and reappeared, moving horizontally to the right with a constant velocity of 8, 14, or 20°/s for a duration of 1,740 ms. The target was then extinguished for 1,200 ms before the start of the next presentation. Subjects were instructed to return toward the start position when the moving target was extinguished. In a further 2 control trials (type II) the target was also extinguished for 420 ms and reappeared, moving horizontally to the right with a constant velocity of 8, 14, or 20°/s for a duration of 660 ms. The target was then extinguished for 1,080 ms, before a second auditory warning cue of 80-ms duration signaled the end of the target motion. During this time the mirror continued to turn at the same rate, and hence the target continued to move, although not visible, with a constant velocity. Subjects were instructed that, although the target would not reappear, they should continue to pursue the target during the nonvisible portion of the trajectory and return to the start position only after hearing the second auditory warning cue. There was a 1,200-ms interval before the start of the next presentation. The visible and nonvisible portions of the trajectory in control trials were balanced such that the overall duration was 4,320 ms.
In experimental trials the target was extinguished for 420 ms and reappeared, moving horizontally to the right with a constant velocity of 8, 14, or 20°/s for a duration of 660 ms. The target was then extinguished for a 420-, 660-, or 900-ms ISI. During the ISI the mirror turned at the same rate and thus the nonvisible target continued to move with a constant velocity. The target was reilluminated and reappeared moving with the same constant velocity for 420 ms (ramp 2). The target was then extinguished for 1,440, 1,200, or 960 ms before the start of the next presentation. The duration of this final part of the presentation was balanced with the duration of the ISI such that each presentation was 4,320 ms. Subjects were instructed to pursue the target during both the visible (ramp 1 and ramp 2) and nonvisible (ISI) portions of the trajectory. An example of representative experimental and control presentations, and the corresponding eye displacement and velocity are shown in Fig. 1.
|
In the control trials the target velocity was blocked for each of the 6 consecutive presentations within that trial (8, 14, or 20°/s). The order in which each velocity was received was randomized across subjects. In 3 of the experimental trials the ISI was arranged in a blocked order (420, 660, or 900 ms), whereas the target velocity was blocked for each of the 6 consecutive presentations within that trial (8, 14, or 20°/s). The order in which each ISI was received was randomized across subjects. In the other 3 experimental trials the ISI and target velocity were randomized as separate factors to avoid possible grouping of presentations that can occur with completely randomized designs. The 3 blocked and randomized trials were received in balanced, randomized combinations to minimize any sequence effects.
Data analysis
Eye velocity and acceleration were derived from eye position using a 2-point central difference algorithm. Eye movements were then analyzed by first identifying and removing saccades from the response using a technique similar to that described previously (Barnes 1982). Saccades were first identified as points in the acceleration trace exceeding a threshold (±2.5 SD of baseline noise: 
750–1,500°/s2). When the threshold criteria were exceeded the complete saccade trajectory was identified by finding the peak and trough of acceleration. On the rare occasions when the use of the acceleration threshold failed to identify a saccade, a second pass was made in which a velocity threshold (40°/s) was applied. Data points equivalent to 8 ms at the beginning and end of the identified saccade trajectory were then excluded to ensure that no saccadic element remained when applying subsequent interpolation. By use of these criteria saccades of 0.3° or more were reliably detected. A linear interpolation routine was used to bridge the gaps produced by removal of saccades from the eye velocity trajectory. Saccades were generally of small amplitude (<5°) and brief duration, making linear interpolation a simple and adequate method of waveform restoration. Data on the saccades during a presentation were stored for later analysis. The desaccaded eye velocity data were then filtered at 35 Hz with a low-pass, zero-phase filter. The resulting eye velocity data were averaged separately for each subject at each level of target velocity (control trials), and target velocity and ISI (experimental trials). The noise (SD) within the position and velocity data during a 500-ms period of static fixation was less than about ±0.1 and ±0.4°/s, respectively.
For the experimental trials the eye velocity at onset of ramp 1 and ramp 2, denoted by V01 and V02, was derived for each subject from the averaged response to the 6 presentations of each combination of the independent variables. These values were examined because they correspond to a time at which the response is considered to be uninfluenced by visual feedback and therefore represent smooth pursuit driven by extraretinal inputs alone. Eye velocity during the ISI was determined at the moment of target offset, and 52, 100, 152, 200, and 252 ms after target offset, respectively. We then extracted the minimum velocity (Vmin) and time of minimum velocity (TVmin) during the ISI. To examine whether there was an increase in eye velocity after Vmin, and whether this was maintained up to the moment of target reappearance (V02), eye acceleration was derived from the slope of a linear regression on eye velocity ±20 ms on either side of this time.
Because smooth pursuit of nonvisible target motion is often combined with saccades (Becker and Fuchs 1985; de Brouwer et al. 2001), we extracted the number of saccades during the ISI. To account for the dynamic overshoot that occasionally follows a saccade (see Lee and Zee 1991), only saccades with absolute amplitude >1° were included in the analysis. Although there was no retinal position and/or velocity error during the ISI, it was decided to examine the accuracy of saccades in bringing the eye back toward the nonvisible target trajectory. To this extent, we calculated the difference between the target displacement and eye displacement at the end of each saccade. We also extracted the number of saccades in the 350-ms interval after target reappearance, and the corresponding average saccade amplitude. We chose to examine a reduced section of the second ramp because it represents sufficient time to use the available visual feedback to make saccadic corrections. Further, it was felt that the final section of the ramp could have included saccades associated with the return to the start point. Observation of the raw data revealed that saccades associated with the return to the start point did not occur during the 350-ms interval after target reappearance.
To determine whether there was any effect of the independent variables on pursuit in the experimental trials the intraindividual means for each dependent variable were submitted to separate 2 condition (random, blocked) by 3 ISI (480, 660, 900 ms) by 3 velocity (8, 14, 20°/s) ANOVA with repeated measures on all factors. Main and interaction effects were further analyzed using Tukey's HSD post hoc procedure. The critical alpha level was set at P < 0.05. Where there were no saccades, the corresponding saccade amplitude for that subject was zero, and there were missing cells in the design. In such cases, missing values were estimated using a regression procedure, and a type IV sum of squares was used in the ANOVA. When the number of missing cells precluded satisfactory estimation, the subject's data were removed from the analysis.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
The regular timing of the auditory cue combined with the constant interval between fixation offset and onset of the first ramp facilitated the generation of anticipatory smooth pursuit before target appearance at ramp 1 (V01). Although there was some between-subject variation, all subjects exhibited eye velocity >1°/s as the moving target first became visible. In the blocked condition, V01 increased significantly with increasing target velocity (Fig. 2) with group means (±SE) of 3.4 ± 0.5, 4.2 ± 0.6, and 5.5 ± 0.8°/s for the 8, 14, and 20°/s targets, respectively. Inspection of the individual subject data revealed this scaling was clearly evident in 7 of the subjects. By contrast, in the randomized condition V01 did not increase linearly with target velocity (Fig. 2). This was a result of V01 to the 8°/s target being higher in the random compared with blocked condition, with group means (±SE) of 5.7 ± 1.1 and 3.4 ± 0.5, respectively. This pattern of results in the group data was evident in all subjects. There was no systematic difference between the random and blocked conditions for the 14 and 20°/s targets. It is likely that the lack of linear scaling in V01 in the random condition was a result of the ensuing target velocity being unpredictable, leading subjects to scale eye velocity according to an average of the 3 possible target velocities.
|
Pursuit during transient target disappearance
After the disappearance of the moving target subjects maintained ongoing pursuit using a combination of saccades and smooth pursuit (Fig. 1). Eye velocity was not maintained at a constant level throughout the ISI, but it did not decay to zero in any of the presentations. All subjects exhibited a significant decrease in eye velocity after target offset. This reduction was more marked when pursuing the 14 and 20°/s targets, but still eye velocity remained scaled to target velocity (Fig. 3). The reduction in eye velocity did not continue for the duration of the ISI. Eye velocity decayed to a minimum (Vmin) that increased as a function of target velocity and was also higher in the blocked than in the random condition when pursuing the 20°/s target. This pattern was evident in each subject, indicating that eye velocity did not decay to a constant, default value (Table 1).
|
|
To examine whether the significant decrease in eye velocity after target offset was simply a result of an underlying oscillation in the pursuit response, we compared Vmin from the experimental trials with velocity at corresponding times (TVmin) in the control trials. The intraindividual mean eye velocity was submitted to separate 3 trial (experimental, controls) by 2 condition (random, blocked) by 3 ISI (480, 660, 900 ms) by 3 velocity (8, 14, 20°/s) ANOVA with repeated measures on all factors. For each target velocity, Vmin was significantly lower than velocity in type I control trials in which the target remained visible throughout its motion. By contrast, there was no significant difference between Vmin and the corresponding velocity in type II control trials in which the target was not expected to reappear. These findings confirm that the decrease in velocity during the ISI was a response to the loss of visual feedback and did not simply reflect oscillations in the normal pursuit response (see Table 1). As expected there were individual differences in the pursuit response during the control trials, but the underlying trend was evident in the majority of subjects (see Figs. 6 and 7).
|
|
In general, after reaching a minimum velocity (Vmin), an increase in eye velocity was made before target reappearance at ramp 2 (Figs. 6 and 7). To determine whether the increase was significant, the intraindividual means of Vmin and V02 were submitted to a 2 condition (random, blocked) by 3 ISI (480, 660, 900 ms) by 3 velocity (8, 14, 20°/s) ANOVA with repeated measures on all factors. This indicated that there was a significant increase between Vmin and V02 across each level of condition, ISI, and velocity (see Fig. 4). This pattern of increasing velocity was particularly evident in 7 of the subjects, especially when pursuing the 14 and 20°/s targets (Figs. 5 and 6). For these subjects, there were only 6 instances out of a possible 84 (2 target velocities, 3 ISIs, 2 conditions, 7 subjects) where Vmin occurred after the end of the ISI, and they were in the 420-ms ISI only. However, Vmin never occurred more than 80 ms after target onset, indicating that the response was not a reaction to the reappearance of the moving target. In the other 2 subjects (5 and 6) there was a less-consistent increase in eye velocity before target reappearance. Eye velocity decreased after the disappearance of the target, but to a somewhat lesser extent, and was then generally well maintained for the remainder of the ISI (Fig. 7). Because of the subjectivity in determining TVmin in the 2 subjects who did not exhibit a clear increase before target reappearance, their data were excluded from the statistical analysis of TVmin. Analysis of the remaining group data (n = 7) revealed no significant influence of condition, ISI, or velocity (Table 1 and Fig. 8). Therefore although there was some variability between subjects, there was no clear trend for TVmin to occur later during the longer ISIs.
|
|
|
To examine whether the difference between the discrete measures, Vmin and V02, was present across the intervening part of trajectory we performed linear regression on the eye velocity data from Vmin to 80 ms after ramp 2 onset. For the group data, the slope of regression equation was 5.2, 6.3, and 7.2 for the 8, 14, and 20°/s targets, indicating that acceleration was reasonably well scaled to target velocity. For the 420- and 660-ms ISIs, excluding the limited examples in which Vmin occurred after target onset at ramp 2, the slope was always positive and significantly different from zero for 7 of the subjects. In the other 2 subjects, the slope was generally closer to zero, and negative in 4 of the possible 12 cases for subject 6 (see Fig. 7). For the 900-ms ISI the slope was generally closer to zero, but it was still predominantly positive and significantly different from zero.
Given that there was some consistency in the timing of the decay of eye velocity after target offset, it was decided to examine the temporal development of individual subjects' mean response after Vmin. To this end, we derived eye velocity at specific times during the ISI before the influence of visual feedback. Eye velocity at 420 ms was compared for the 420-, 660-, and 900-ms ISI, and at 660 ms for the 660- and 900-ms ISI. ANOVA indicated that, whereas eye velocity at 420 and 660 ms after target offset increased significantly with target velocity, it did not change as a function of ISI (Fig. 9 and Fig. 10). There was a significant interaction between condition and target velocity, with 7 of the subjects exhibiting a higher velocity at both 420 and 660 ms after target offset in the blocked compared with random condition when pursuing the 20°/s target. This analysis indicates that the temporal development of eye velocity was similar across the different ISIs, and that the magnitude at the discrete key moments was influenced by condition and target velocity alone.
|
|
It was also found that there were more saccades in the 900-compared with 420- and 660-ms ISIs, and when pursuing the 14 and 20°/s targets compared with the 8°/s target. In other words, the number of saccades increased as a function of both ISI and target velocity. Given that there was no retinal position and/or velocity error, the saccades (n = 561) were reasonably successful in bringing the eye toward the nonvisible target trajectory (Fig. 11A). Although there was a tendency to overshoot the target (357) there was a significant linear relation between eye and target displacement at the end of each saccade (R2 = 0.73). This is evident in Fig. 11B, which shows the distribution of the difference between the eye and target displacement. Of the total distribution, 72% of saccades brought the eye back within ±1.5° of the target.
|
Pursuit at and after target reappearance
The influence of condition and target velocity found during the ISI was still apparent at the moment of target reappearance. V02 was higher when pursuing the 20°/s target in the blocked compared with the random condition, across all ISIs. This was particularly evident in 6 of the subjects. For the group, there was no difference in V02 between the blocked and random conditions for the 8 and 14°/s targets, across each level of ISI. For 6 of the subjects, V02 was very similar in the 2 conditions, whereas for the other 3 subjects there was no particular trend. Regardless of the absolute level of velocity, V02 generally increased as a function of target velocity for both blocked and random conditions (Fig. 12).
|
As noted above, although V02 was significantly higher than Vmin across all levels of target velocity, there was some evidence that the eye did not continue to accelerate up to and beyond the moment of target reappearance. When pursuing the 20°/s target the eye was accelerating only at the end of the 420- and 660-ms ISIs. As can be seen in Fig. 5D and Fig. 6 (bottom panel; see also Table 3), the eye was decelerating as the target reappeared after a 900-ms ISI. Each subject exhibited this effect. For the 8 and 14°/s targets there was more intersubject variation, but still eye acceleration was close to zero as the target reappeared after the 900-ms ISI. Because the eye was decelerating in some cases, V02 did not necessarily represent the peak level of eye velocity (Vpk) attained after Vmin. The intraindividual means of Vpk and V02 were submitted to a 2 condition (random, blocked) by 3 ISI (420, 660, 900 ms) by 3 velocity (8, 14, 20°/s) ANOVA with repeated measures on all factors. This indicated that there was a significant difference between Vpk and V02 for the 20°/s target only. Vpk more closely approximated target velocity than V02. This was particularly evident when pursuing the 20°/s target across the 900-ms ISI, where the group mean Vpk was 15.6 and 17.5°/s in the random and blocked conditions, and the corresponding V02 values were 13.6 and 14.8°/s, respectively. The influence of condition on eye velocity at the moment of target reappearance was generally well reflected in the data on the amplitude of corrective saccades. Subjects exhibited larger-amplitude corrective saccades after target reappearance when pursuing the 20°/s target in the random compared with the blocked condition (Table 2). Although there was no interaction with the duration of the ISI, there was a trend for the amplitude of corrective saccades to increase with ISI. For example, observation of the individual subject data indicated that 8 of the 9 subjects exhibited larger-amplitude corrective saccades after the 900-ms ISI when pursuing the 20°/s target in the random compared with blocked condition. A similar trend was evident in 7 of the 9 subjects after the 660-ms ISI. There was no influence of condition on the number of saccades in the 350 ms after target reappearance. Similar to when the target was not visible during the transient, the number of saccades increased as a function of both ISI and target velocity. However, as expected given that there was retinal position and velocity information available, subjects rarely exhibited more than one corrective saccade.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
; Mitrani and Dimitrov 1978), but can be sustained with a reduced gain if subjects exert volitional effort to maintain pursuit (Becker and Fuchs 1985; Pola and Wyatt 1997). Depending on the duration for which the moving target is nonvisible, it inevitably follows that there will be a velocity error at the moment of target reappearance. However, this could be reduced if eye velocity increases back toward target velocity to coincide with the moment of target reappearance. In the present study, we confirmed that eye velocity was not maintained close to target velocity for the duration of the transient disappearance. In the majority of subjects (n = 7), eye velocity did not decay to a plateau that was maintained until the target reappeared. Rather, eye velocity decayed until it reached a minimum, which generally occurred before the target reappeared, and then increased back toward target velocity. This was particularly evident for the 14 and 20°/s targets. Even in the limited examples in which the minimum occurred after target reappearance (420-ms ISI only), this was within 100 ms, indicating that the subsequent increase in eye velocity was not simply a reaction to visual feedback. Rather, the predominant trend was for the generation of an increase in eye velocity before target reappearance. We found no evidence of a similar pattern in the eye velocity data measured in control trials. These findings indicate that the decrease and subsequent increase in eye velocity during the ISI was influenced by the result of the experimental manipulations, and was therefore anticipatory in the sense that it was dependent on expectation of target reappearance. As expected, the temporal development of eye velocity during the ISI was relatively variable between subjects. Still, there were some general trends that are worth consideration. A notable feature was that TVmin did not increase significantly as a function of ISI in either the random or blocked presentations. Subjects generated the increase in eye velocity at a comparatively similar time (TVmin, mean = 359, SE ±8 ms), over the range of ISIs. The increasing velocity trajectory after TVmin was also very similar within subjects, irrespective of ISI. This conclusion was supported by the observation that eye velocity was not significantly different at 420 ms for all ISIs, and at 660 ms for ISIs of 660 and 900 ms. As a consequence of the stereotypical response, the increase in eye velocity occurred too early in the 900-ms ISI. This was particularly evident when pursuing the 20°/s target, and culminated in the eye decelerating up to and beyond the moment of target reappearance, until visual feedback became available. Thus although apparently anticipatory, the timing of the response was not appropriate to the duration of the ISI.
At first sight, this observation appears unexpected, given the previous finding that anticipatory smooth pursuit can be initiated with fairly precise timing to repeated presentations of predictable stimuli (Barnes and Donelan 1999; Kao and Morrow 1994). It would seem likely that the repetition of stimuli with identical velocity and temporal characteristics, as in the blocked condition, would result in the time between initiation of anticipatory eye movement and target reappearance being fairly constant across the range of ISIs. It is possible that having received only 6 repeated presentations, subjects were not able to use the information contained in the identical presentations to derive a precise estimate of when to increase eye velocity. Rather, subjects may have used a timing estimate based on experience of previous presentations of the different target velocities and ISIs. This estimate may have been biased toward the shorter duration ISI because it would ensure that eye velocity was more appropriate at target reappearance, even if it was decaying, for the majority of presentations. An alternative position is that, unlike the initiation of anticipatory smooth pursuit from a stationary location, the timing of the increase in eye velocity during the ISI is not predictive of the target's reappearance. An increase in eye velocity at a fixed time after target offset could have been triggered if it decayed to a particular threshold level, which took a similar time to reach with the different target velocities. However, the finding that this was not evident when the expectancy was that the target would not reappear indicates that, regardless of the mechanism controlling this response, it was anticipatory of the target's future motion. To determine whether the increase during the transient can be influenced by prediction, further work is necessary to examine whether subjects learn to time the increase in eye velocity such that it coincides more appropriately with target reappearance, and whether the provision of more salient timing cues, such as those that can be interpreted from an occluding surface, influence this process.
Our argument so far has assumed that the increasing eye velocity after TVmin is linked in some way with anticipatory behavior. However, it could also be argued that it simply reflects the oscillatory dynamics of normal sustained pursuit. There are 2 reasons for rejecting this interpretation. First, a comparison with control data indicated that there was not a similar decrease in eye velocity when the target remained visible throughout the presentation. Subjects maintained pursuit with a gain close to unity in these control trials, and did not exhibit low-frequency, large-amplitude oscillations. Second, if viewed as oscillatory behavior, the frequency of the response during the ISI in experimental trials was about 1 Hz (e.g., Fig. 6, bottom panel), which is far lower than the 3- to 4-Hz oscillations previously observed in the presence of the visible target (Goldreich et al., 1992; Robinson et al., 1986). Moreover, this oscillatory behavior is still overlaid on an increasing baseline, as indicated by the (mostly) positive slopes of the regression between TVmin and ramp 2 onset. An alternative account of our results is that the measured response was simply a nonanticipatory recovery (i.e., the onset of a low-frequency oscillation) from a transient loss of drive attributed to the removal of visual feedback. This is not consistent with our control data, which showed that subjects did not exhibit any increase in eye velocity after target offset when there was no expectation of reappearance. It is also not consistent with the previous finding that even over longer ISIs (e.g., 1,620 or 4,000 ms) eye velocity tends to decay exponentially, without a recovery until the target reappears (Barnes and Asselman 1991, 1992; Becker and Fuchs 1985) or the end of the presentation is reached (Pola and Wyatt 1997). If the decrease and subsequent increase in eye velocity constituted a nonanticipatory, low-frequency oscillation after the disappearance of the target, this should be evident whenever there is a loss of visual feedback. This is clearly not the case, and in fact to our knowledge has not been previously documented. Therefore although we do not discount the possibility that the increase in velocity could reflect the onset of oscillation after a loss of drive, we suggest that it is dependent on the expectation of target reappearance and thus a form of anticipatory behavior.
A model of ocular pursuit to the transient disappearance of a moving visual target
Several authors have proposed models of ocular pursuit in which the drive for eye velocity can be influenced by both retinal and extraretinal inputs (Barnes et al. 1995; Krauzlis and Lisberger 1994; Krauzlis and Miles 1996; Leigh and Zee 1991; Robinson et al. 1986). Models of this basic architecture can simulate the effect of stabilizing the target image on the retina. Although visual feedback signals zero error, there is a continued contribution from an efference copy of the motor drive, resulting in a reduced eye velocity (Pola and Wyatt 1997). When a moving target disappears and the subject is left in darkness, there is no visual feedback signal confirming the output of the efference copy reafferent feedback system. To simulate this it has been suggested that the lack of a visual feedback signal inhibits the efference copy by setting its gain to zero such that once it has been released and played out, all further output is inhibited and eye velocity decays exponentially to zero (Barnes et al. 1995). There is also evidence that the effect of an absent visual feedback signal can be somewhat overcome by volitional effort, prolonging the response (Becker and Fuchs 1985; Pola and Wyatt 1997). This can be simulated by reinstating the gain of efference copy after a short interval. However, because the efference copy loop functions like an integrator, once its feedback is reduced, it cannot be reinstated at the original level unless its gain is set higher than unity, which would potentially create instability (Dallos and Jones 1963). Thus it is very difficult to account for the subsequent increase in eye velocity back toward its previous value that we have observed in the current experiment.
To generate eye velocity with an increasing profile up to target reappearance without increasing gain beyond unity, it is necessary that the visuomotor drive is not influenced by changes within the efference copy feedback. Below we present an alternative model that incorporates this criterion and which offers behaviorally realistic simulations and compatibility with other findings. In this model the drive for eye velocity can also be influenced by both retinal and extraretinal inputs. Following on from Barnes and Asselman (1991) and Barnes and Wells (1999), the extraretinal input has been modeled as an internal reafferent feedback system, which can be influenced by expectation, volition, attention, and visual feedback regarding the consequences of the ongoing ocular response. The term reafferent is used because the feedback is of a postsensory but premotor drive, and not an efference copy of the eye movement itself (Barnes et al. 1997). The reafferent feedback system is arranged to allow the temporary creation of a short-term store that represents velocity-coded information. Figure 13 depicts a simplified version of this model in which it is assumed that reafferent feedback inputs to a short-term store (MEM). MEM has been represented as a local feedback loop, similar to that proposed for the saccadic system (Robinson 1975). It contains an integrator that summates the error within the local feedback loop until the error is zero (NB: this is simplified for unidirectional movement). The output of this loop thus reaches a level equivalent to the visuomotor drive (vmd). In effect, it acts as a sample-and-hold mechanism. One consequence of this arrangement is that the gain 
of the reafferent feedback need not be unity to maintain the store. Therefore if there is a temporary modification in the stored level of the reafferent feedback will remain the same. This contrasts with other models in which the output of the integrator is influenced by a reduction in gain, causing a reduction in velocity that is not recovered to its previous level without any further visual input (e.g., Krauzlis and Miles 1996). A further advantage of our model is that it is also consistent with results from previous experiments examining the generation of anticipatory eye movements, the scaling of which are dependent on stored information obtained from prior stimulation. For example, it can account for the finding that anticipatory smooth pursuit is not eliminated by several seconds of fixation on a stationary target between presentations of moving stimuli (Barnes et al. 1997; Chakraborti et al. 2002), whereas a conventional efference copy store would be discharged when eye velocity went to zero during fixation.
|
Using this model, the response observed when the target disappears can be simulated by temporarily reducing gain applied within the reafferent feedback loop. If is reduced from its normal value (0.95) to zero for a short period (300 ms in Fig. 14A) eye velocity will decay to a minimum, but then recover toward target velocity, as in the majority of our responses. It recovers more slowly than it decays because the output of MEM is fed through a low-pass filter that was originally included specifically to simulate anticipatory smooth pursuit (Barnes 1994). If goes to zero for sufficient time eye velocity will decay exponentially to zero (Fig. 14C). To simulate the maintenance of constant eye velocity over longer ISIs (Becker and Fuchs 1985; Pola and Wyatt 1997) it is necessary to reduce to an intermediate value, rather than to zero. By reducing , eye velocity can be maintained at a reduced level for a brief period (Fig. 14B) or for an indefinite duration (Fig. 14D), before increasing to its previous level when is reinstated to its normal level. Between-subject variability in the rate and magnitude of the decay would then occur because of individual differences in the level of visuomotor drive and the intermediate level of assumed. Previous work indicates that the decay will be influenced by the ocular response before target disappearance and thus is a complex function of ramp duration, target velocity, number of presentations, and expectation (Becker and Fuchs 1985; Ohashi and Barnes 1996). This is consistent with our finding that presentation order (random vs. blocked) influenced the magnitude of eye velocity from TVmin up to the moment of target reappearance.
|
The data from 7 of our subjects could equally support the idea that gain is reduced to zero or some intermediate value. Both approaches would generate a relatively stereotypical development of eye velocity during the transient if gain were modified at a similar time regardless of the duration of the ISI. However, the finding that 2 of our subjects maintained pursuit at a reduced level throughout the ISI is better simulated by reducing gain to an intermediate value (cf. Becker and Fuchs 1985; Pola and Wyatt 1997). To appropriately time the increase in eye velocity by modification of gain , one should ideally take account of the ISI and predict the onset of target motion. Our subjects did not appear able to use the predictable information contained in the blocked presentations to modify their response. Factors such as expectation, attention, and experience of prior stimuli could influence this process. However, as noted earlier, the modification of gain may not be predictive of target reappearance time, but rather triggered when eye velocity reaches a threshold level. Interestingly, the 2 subjects who maintained pursuit at reduced velocity throughout the ISI also generally exhibited the highest Vmin. In these subjects eye velocity may not have reached the level necessary to trigger an increase by a modification of gain . At present it is an open question how this is implemented in our model.
The modification of gain can also explain how subjects learn to respond to intermittently presented targets constituting more complex sequences such as multiple different velocity ramps. Barnes and colleagues (Barnes and Asselman 1992; Barnes and Schmid 2002) previously suggested that subjects pursue complex sequences by releasing separate predictive estimates in anticipation of target reappearance time and velocity (see also Boman and Hotson 1992). Having assumed that gain was reduced to zero between estimates, they demonstrated that by summating responses to individual stimulus ramps, it was possible to simulate a response that was continuously predictive. As we discuss here, the reduction of gain to zero does not enable the response to be sustained over longer interramp intervals. Therefore we suggest that the response is most likely a continuous prediction formed by modifying gain within the reafferent feedback system at discrete intervals. Also, rather than storing several levels of velocity-coded information and releasing the related response at the appropriate time, grading the gain controller can generate a similar response. To reduce the potential instability of increasing gain beyond unity, we suggest that gain would be graded relative to the highest target velocity in the sequence. Findings using either single ramp (Jarrett and Barnes 2002) or multiple ramp stimuli (Barnes and Schmid 2002) indicate that prior exposure enables at least 4 levels of velocity to be graded in this way.
Although our aim was not to provide an exhaustive description of the possible means of generating eye velocity with an increasing profile up to target reappearance, an alternative that is worthy of further consideration would be to modify the open-loop gain K of the system downstream of the reafferent loop (Fig. 13). This would preserve the stored information within the reafferent loop, such that an anticipatory increase could be generated before target reappearance, by reinstating gain after a temporary reduction. As desired, this produces a response that decays rapidly and then plateaus at a reduced level before rising in anticipation of target reappearance. However, it generates an anticipatory response during the ISI that has an abrupt acceleration that does not match closely the measured slower increase in eye velocity before target reappearance (Fig. 7; see also Barnes and Schmid 2002).
In summary, although subjects' extrapolated smooth pursuit over a period of nonvisible target motion, they did not maintain eye velocity close to target velocity, particularly with the 14 and 20°/s targets. In response to the reduction in eye velocity, most subjects released a further increase in eye velocity before the onset of the second ramp. This increase was not precisely timed to the different times of target reappearance and thus the response occurred too early in the longer ISI. There was no evidence of a similar increase when subjects knew the target would not reappear. We suggest that the response was anticipatory in the sense that it was dependent on expectation of the target's future motion. We provide a model in which these effects are explained by the modification of gain within a reafferent feedback system containing a short-term store.
|
DISCLOSURES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: S. J. Bennett.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESREFERENCES |
|---|
Barnes GR and Asselman PT. The mechanism of prediction in human smooth pursuit eye movements. J Physiol 439: 439–461, 1991.
Barnes GR and Asselman PT. Pursuit of intermittently illuminated moving targets in the human. J Physiol 445: 617–637, 1992.
Barnes GR, Barnes DM, and Chakraborti SR. Ocular pursuit responses to repeated, single-cycle sinusoids reveal behavior compatible with predictive pursuit. J Neurophysiol 84: 2340–2355, 2000.
Barnes GR and Donelan AS. The remembered pursuit task: evidence for segregation of timing and velocity storage in predictive oculomotor control. Exp Brain Res 129: 57–67, 1999.[ISI][Medline]
Barnes GR, Donnelly SF, and Eason RD. Predictive velocity estimation in the pursuit reflex response to pseudo-random and step displacement stimuli in man. J Physiol 389: 111–136, 1987.
Barnes GR, Goodbody SJ, and Collins S. Volitional control of anticipatory ocular pursuit responses under stabilized image conditions in humans. Exp Brain Res 106: 301–317, 1995.[ISI][Medline]
Barnes GR, Grealy MA, and Collins S. Volitional control of anticipatory ocular smooth pursuit after viewing, but not pursuing, a moving target: evidence for a reafferent velocity store. Exp Brain Res 116: 445–455, 1997.[ISI][Medline]
Barnes GR and Schmid AM. Sequence learning in human ocular smooth pursuit. Exp Brain Res 144: 322–335, 2002.[ISI][Medline]
Barnes GR and Wells SG. Modelling prediction in ocular pursuit: the importance of short-term storage. In: Current Oculomotor Research: Physiological and Psychological Aspects, edited by Becker W, Deubel H, and Mergner T. New York: Plenum, 1999, p. 97–107.
Becker W and Fuchs AF. Prediction in the oculomotor system: smooth pursuit during transient disappearance of a visual target. Exp Brain Res 57: 562–575, 1985.[ISI][Medline]
Boman DK and Hotson JR. Predictive smooth pursuit eye movements near abrupt changes in motion direction. Vision Res 32: 675–689, 1992.[ISI][Medline]
Dallos P and Jones R. Learning behaviour of the eye fixation control system. IEEE Trans Autom Contr AC 8: 218–227, 1963.
de Brouwer S, Missal M, and Lefèvre P. Role of retinal slip in the prediction of target motion during smooth pursuit. J Neurophysiol 86: 550–558, 2001.
Goldreich D, Krauzlis RJ, and Lisberger SG. Effect of changing feedback delay on spontaneous oscillations in smooth pursuit eye movements of monkeys. J Neurophysiol 67: 625–638, 1992.
Heinen SJ and Watamaniuk SNJ. Spatial integration in human smooth pursuit. Vision Res 38: 3785–3794, 1998.[ISI][Medline]
Jarrett CB and Barnes GR. Volitional selection of direction in the generation of anticipatory smooth pursuit in humans. Neurosci Lett 312: 25–28, 2001.[ISI][Medline]
Kao GW and Morrow MJ. The relationship of anticipatory smooth eye movement to smooth pursuit initiation. Vision Res 34: 3027–3036, 1994.[ISI][Medline]
Kowler E and Steinman RM. The effect of expectations on slow oculomotor control. II. Single target displacements. Vision Res 19: 633–646, 1979.[ISI][Medline]
Krauzlis RJ and Lisberger SG. Temporal properties of visual motion signals for the initiation of smooth pursuit eye movements in monkeys. J Neurophysiol 72: 150–162, 1994.
Krauzlis RJ and Miles FA. Transitions between pursuit eye movements and fixation in the monkey: dependence on context. J Neurophysiol 76: 1622–1638, 1996.
Leigh RJ and Zee DS. The Neurology of Eye Movements. Philadelphia, PA: Davis, 1991.
Lyon DR and Waag WL. Time course of visual extrapolation accuracy. Acta Psychol 89: 239–260, 1995.[Medline]
Mitrani L and Dimitrov G. Pursuit eye movements of a disappearing moving target. Vision Res 18: 537–539, 1978.[ISI][Medline]
Morris EJ and Lisberger SG. Different responses to small visual errors during initiation and maintenance of smooth-pursuit eye movements in monkeys. J Neurophysiol 58: 1351–1369, 1987.
Ohashi N and Barnes GR. A comparison of predictive and non-predictive ocular pursuit under active and passive stimulation conditions in humans. J Vestib Res 6: 261–276, 1996.[ISI][Medline]
Peterken C, Brown B, and Bowman K. Predicting future position of a moving target. Perception 20: 5–16, 1991.[Medline]
Pola J and Wyatt H. Offset dynamics of human smooth pursuit eye movements: effects of target presence and subject attention. Vision Res 37: 2579–2595, 1997.[ISI][Medline]
Robinson DA. Oculomotor control signals. In: Basic Mechanisms of Ocular Motility and Their Clinical Implications, edited by Lennerstrand F and Bach-y-Rita P. Oxford, UK: Pergamon, 1975, p. 337–374.
Robinson DA, Gordon JL, and Gordon SE. A model of the smooth pursuit eye movements system. Biol Cybern 55: 43–57, 1986.[ISI][Medline]
Tanaka M and Fukushima K. Slow eye movement evoked by sudden appearance of a stationary visual stimulus observed in a step-ramp smooth pursuit task in monkey. Neurosci Res 29: 93–98, 1997.[ISI][Medline]
Tanaka M and Lisberger SG. Context-dependent smooth eye movements evoked by stationary visual stimuli in trained monkeys. J Neurophysiol 84: 1748–1762, 2000.
This article has bee
/2)–0.5; visual feedback delay: 
v = 0.1 s; KM = 10; K = 2.8. Mean response output represents output from MEM.