Free
Article  |   December 2012
Facilitation of ocular pursuit during transient occlusion of externally-generated target motion by concurrent upper limb movement
Author Affiliations
Journal of Vision December 2012, Vol.12, 17. doi:10.1167/12.13.17
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Simon J. Bennett, Daniel O'Donnell, Steve Hansen, Graham R. Barnes; Facilitation of ocular pursuit during transient occlusion of externally-generated target motion by concurrent upper limb movement. Journal of Vision 2012;12(13):17. doi: 10.1167/12.13.17.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract
Abstract
Abstract:

Abstract  Smooth pursuit during prolonged occlusion is improved in the presence of sensorimotor signals when tracking self-generated target motion. The current study investigated whether concurrent arm tracking of externally-generated target motion conveys a similar facilitation to ocular pursuit of transiently occluded constant velocity (Experiment 1) or accelerating (Experiment 2) targets. Velocity characteristics and occlusion duration were arranged in random or blocked order, thus permitting a novel examination of the contribution from sensorimotor signals and predictive processes acting within the ocular system during transient occlusion. Consistent with previous investigations, smooth pursuit decayed during transient occlusion; but eye velocity was higher when trials were presented in blocked compared to random order, particularly for positively accelerating targets. For fast, constant velocity targets, concurrent arm movement facilitated smooth pursuit during transient occlusion. Nevertheless, even with increased predictability regarding the upcoming target motion in blocked-order trials and the presence of sensorimotor signals from concurrent arm movement, eye velocity always remained less than target velocity during occlusion. This contrasted with the manual response, which attained velocity close to target velocity, whether in blocked or random conditions. These findings are discussed with reference to recent models of ocular pursuit that incorporate short-term and/or long-term prediction to account for target extrapolation during occlusion.

Introduction
Improvements in control of smooth pursuit (i.e., increased maximum velocity and gain, reduced latency at onset) have been reported in a number of studies where target motion is the result of (i.e., self-generated) concurrent upper limb movement (Gauthier, Vercher, Ivaldi, & Marchetti, 1988; Steinbach, 1969; Vercher, Quaccia, & Gauthier, 1995). It has been suggested that the increased magnitude of smooth pursuit can be attributed to limb afference (Gauthier & Hofferer, 1976; Gauthier & Ivaldi, 1988), whereas limb efference is thought to be mainly involved in reducing latency (Vercher et al., 1996). Compared to eye-alone tracking, the contribution of sensorimotor signals from cyclical forearm or finger movement has been shown to be particularly beneficial in maintaining gain of smooth pursuit during prolonged absence of visual feedback (Gauthier & Hofferer, 1976; Gauthier & Ivaldi, 1988; Gauthier et al., 1988). For instance, when instructed to stop moving the limb (i.e., forearm) to which an imagined target was attached, smooth pursuit decayed to zero in approximately 600 ms, after which eye movements became almost exclusively saccadic for the remainder of the trial (Gauthier & Hofferer, 1976). Smooth pursuit was resumed when participants were subsequently instructed to start moving their limb, although initially at lower gain and with more corrective saccades. 
While not intending to question the proposed role of sensorimotor signals in putative models of oculo-manual coupling (coordination control system) (Lazzari, Vercher, & Buizza, 1997; Vercher, Lazzari, & Gauthier, 1997), it is notable that the reported difficulty maintaining smooth pursuit in the absence of visual and proprioception feedback (e.g., following termination of self-generated target motion in darkness) is much more severe than when tracking externally-generated target motion. For instance, when presented with a range of short duration occlusions (e.g., 400-1620 ms) of externally-generated target motion received in random order, smooth pursuit is maintained at reduced gain (Becker & Fuchs, 1985; Pola & Wyatt, 1997) or often increases in anticipation of target reappearance (Bennett & Barnes, 2003; Bennett & Barnes, 2005). Moreover, given just a few trials received in blocked order, both eye displacement and velocity can be scaled such that they reflect the target trajectory at the moment of reappearance (Bennett & Barnes, 2004; Bennett, Orban de Xivry, Barnes, & Lefèvre, 2007; Bennett, Orban de Xivry, Lefèvre, & Barnes, 2010; Orban de Xivry, Bennett, Lefèvre, & Barnes, 2006). Recent modeling can simulate well the ocular response during short-duration transient occlusion by incorporating short-term (i.e., within-trial) and long-term (between-trial) predictive influences in the form of a direct (e.g., efference copy) loop that operates during random-order trials, and an indirect (i.e., internal memory structure) loop that provides more persistent input during blocked-order trials (Ackerley & Barnes, 2011; Bennett et al., 2010). 
Clearly, contrary to traditional belief, the oculomotor system is reasonably capable of extrapolating occluded externally-generated target motion. Indeed, even when the recovery in smooth pursuit during occlusion does not entirely eliminate retinal slip as the target reappears, eye positional control is well maintained (in blocked-order trials) by the contribution of predictive saccades. The implication, therefore, is that previous comparisons of oculo-manual and ocular pursuit of imagined self-generated target motion (Gauthier & Hofferer, 1976) may somewhat misrepresent the advantage afforded by access to concurrent sensorimotor signals. Notably, in said study, the use of a within-trial transition from oculo-manual to ocular pursuit could have resulted in participants having difficulty shifting attention away from forearm movement to an imagined target motion, and thus maintaining the internal prediction of target motion that is critical to drive eye movements in the absence of visual feedback (Barnes & Marsden, 2002). Also, the contribution of sensorimotor signals to smooth pursuit in the absence of visual feedback was much greater when performing cyclical forearm as opposed to finger movement. As recognized by Gauthier and Hofferer (1976), facilitation of smooth pursuit by sensorimotor signals is “related to the complexity or the nature of the self-moved structure” (pp. 134), which was likely optimized when pursuing an imagined target on the forearm (e.g., spatially congruent, large amplitude, familiar movement). 
The current study examined the contribution of sensorimotor signals from the upper limb to ocular pursuit (displacement and velocity) during occlusion of externally-generated target motion. To this end, we compared separate conditions in which participants pursued an externally-generated constant velocity (Experiment 1) or constant acceleration (Experiment 2) ramp with eyes alone (ocular) or eyes and upper limb (oculo-manual). A discrete target motion was visible throughout a presentation (control trials) or either side of occlusion (experimental trials). The occlusion duration was sufficiently short to eliminate the need for movement reversals but long enough (600 or 1000 ms) to elicit an exclusively saccadic ocular response. Using this design, we expected participants would experience the characteristic reduction in eye velocity during occlusion, but less so when sensorimotor signals were available during oculo-manual pursuit. Specifically, it was anticipated that facilitation by sensorimotor signals would be most evident when pursuing fast moving constant velocity or positively accelerating targets as these are particularly difficult to pursue without visual feedback (Bennett et al., 2010). By using externally-generated target motion, we were also able to arrange the upcoming velocity characteristics and occlusion duration in random or blocked order, thus affording short-term and/or long-term prediction (Deno, Crandall, Sherman, & Keller, 1995). The manipulation of upcoming velocity was not possible in studies on pursuit of self-generated target motion, where the participant always had access to sensorimotor signals that could inform the ocular control system. Also, by performing the task in darkness, participants in previous studies had no expectation that the target would be visible later in its trajectory, which is known to be important in maintaining ocular pursuit (Mitrani & Dimitrov, 1978). Accordingly, we have been able to conduct a novel examination of the potential interaction between the contribution from sensorimotor signals and predictive processes acting within the ocular system during transient occlusion. 
Method
Participants
Nine participants for Experiment 1 (mean age 26 ± 7 years) and twelve participants for Experiment 2 (mean age 27 ± 6 years) from the staff and student population at Liverpool John Moores University volunteered to take part in the study. All participants were right handed, had normal or corrected-to-normal vision, were healthy and without any known oculomotor abnormalities. The investigation was conducted according to a protocol approved by Liverpool John Moores University ethics committee in conformity with the tenets of the Declaration of Helsinki. All participants gave written informed consent before taking part in the experiment. 
Apparatus
Participants sat in a purpose-built dark laboratory, facing a flat white screen (2.05 m × 1.54 m) at a viewing distance of 2.01 m; at this distance disconjugate eye movements are minimized. The head was supported by a height-adjustable chin rest and head support, which were placed under the chin and at the nape of the neck, respectively. Participants' eye height from the floor was measured and entered into the stimulus generation routine, allowing for the projected image to appear at eye level directly in front of the subject in the fronto-parallel plane. Visual stimuli were projected onto the screen using a CRT projector (Barco Graphics 908) with a refresh rate of 85 Hz and 1024 × 768 spatial resolution. The visual stimuli were generated on a host PC (Dell Precision 670) using the COGENT toolbox implemented through MATLAB (Mathworks Inc.). 
Movement of the left eye was recorded at 200 Hz using a Chronos 3D eye-tracking device (Chronos Vision). An output signal from the parallel port of the host PC was used to synchronize stimulus generation with eye movement recording. Movement of the right arm was recorded on the host PC in MATLAB (85 Hz) using a hand-held Logitech G5 laser mouse (800DPI; 1000 reports/s). Movement of the mouse was restricted to the (horizontal) x-axis using a metal rail and a low friction slider. Small wooden stoppers located 340 mm apart at the extremes of the rail ensured participants could not remove the mouse from the rail during testing. Before each testing session, the metal rail was aligned with the subject's sternum to comfortably allow for a range of motion up to the maximum linear amplitude of 160 mm (i.e., 40 deg on the screen) required by the stimulus characteristics. 
Procedure
During testing, participants were tasked with tracking a horizontally moving target as accurately as possible with either the eyes (ocular condition) or eyes and arm (oculo-manual condition) in a series of experimental (discrete occlusion) and control (continuously visible) trials. The lights in the room were extinguished throughout testing, thus ensuring that only the targets projected on the screen could be seen. Trials began with a green circular target (0.5° diameter) appearing at −20° left from the screen centre for a fixed duration of 2600 ms (Figure 1). While remaining stationary, the target color changed to red, signaling to the participant that it would soon begin to move. After a random duration between 1650 and 1850 ms the target disappeared for 300 ms before reappearing moving horizontally to the right. In Experiment 1, the target moved with a constant velocity of 10, 15, or 20 deg/s, whereas in Experiment 2 the target moved with constant acceleration of −12, −8, 0, 8, or 12 deg/s2.In experimental trials, the moving target was visible for the initial 600 ms of the trajectory, after which it was occluded (Experiment 1: 600, 800 or 1000 ms; Experiment 2: 600 or 1000 ms). For the 5 levels of target acceleration, initial target velocity was 19.2, 16.8, 12, 7.2, or 4.8 deg/s respectively, thus resulting in target velocity of 12 deg/s at the moment of occlusion, and a change in velocity that was above the suggested perceptual threshold of 25% (Brouwer, Brenner, & Smeets, 2002). During occlusion the target continued to move to the right, with the same velocity and/or acceleration, albeit invisible to the participant. At the end of occlusion the target reappeared moving horizontally to the right for a further 400 ms. Participants were instructed to pursue the target through both the visible and occluded portions of the trajectory. A black screen was presented before the start of the next trial. 
Figure 1
 
Schematic diagram of an experimental trial in the ocular (panel A) and ocular-manual (panel B) conditions. A stationary green target (box 1) was presented for 2600 ms on a black screen before changing color to red (box 2). In ocular-manual conditions, an unfilled white circle represented the motion of the hand-held mouse. After a random period between 1650 and 1850 ms, the red target then disappeared for 300 ms (box 3) and reappeared moving horizontally to the right for 600 ms (box 4). The target was then occluded for 600, 800, or 1000 ms (box 5). The target reappeared moving to the right for 400 ms (box 6), and was then extinguished leaving a blank screen (box 7) before the start of another trial. White arrows represent direction when the target was in motion and were not visible to the subject at any part of the trial.
Figure 1
 
Schematic diagram of an experimental trial in the ocular (panel A) and ocular-manual (panel B) conditions. A stationary green target (box 1) was presented for 2600 ms on a black screen before changing color to red (box 2). In ocular-manual conditions, an unfilled white circle represented the motion of the hand-held mouse. After a random period between 1650 and 1850 ms, the red target then disappeared for 300 ms (box 3) and reappeared moving horizontally to the right for 600 ms (box 4). The target was then occluded for 600, 800, or 1000 ms (box 5). The target reappeared moving to the right for 400 ms (box 6), and was then extinguished leaving a blank screen (box 7) before the start of another trial. White arrows represent direction when the target was in motion and were not visible to the subject at any part of the trial.
In the oculo-manual condition, a cursor that represented motion of the hand-held mouse was projected in combination with the moving target. The cursor was an annulus of 0.8 deg diameter (Figure 1 panel B) and had a gain relationship such that 4 mm of mouse movement on the handrail equated to 1 deg of movement on the screen. The target and cursor were aligned horizontally at the beginning of each trial by the participant, who was tasked with keeping the moving target within the boundaries of the cursor as accurately as possible. Neither the cursor nor the moving target was visible during occlusion in experimental trials. Participants also completed the ocular and oculo-manual tracking in control trials, which were identical to experimental trials except the target, and cursor in the oculo-manual condition, remained visible throughout the trajectory. 
Before commencing with experimental testing, participants visited the laboratory for a familiarization session, during which they completed a single block of random order oculo-manual trials that comprised all combinations of the independent variables. Thiswas important as it allowed participants to experience the stimuli and also adapt to the transformation required to match sensorimotor information when moving the mouse to the visual consequences on the screen. Participants then visited the laboratory on four separate occasions, during which they completed a session of blocked order and random order trials in each of the ocular and oculo-manual conditions. To minimize any sequence effects, the four sessions were received in a different order across participants. In each session of Experiment 1, participants performed three blocks of 36 trials (n = 108), comprising six repeats of each of the following: 2 trial type (control, experimental), 3 target velocity (10 deg/s, 15 deg/s, 20 deg/s), and 3 occlusion duration (600 ms, 800 ms, 1000 ms). In each session of Experiment 2, participants performed 4 blocks of 30 trials (n = 120), comprising 6 repeats of each of the following: 2 trial type (control, experimental), 5 target acceleration (−12, −8, 0, 8 or 12 deg/s2), and 2 occlusion duration (600 ms, 1000 ms).In total, participants completed 432 trials in Experiment 1 and 480 trials in Experiment 2. During blocked-order trials, identical target parameters were presented for 6 consecutive trials. The order in which participants received each combination of target parameters was pseudo-randomized. For random-order trials, target parameters were pseudo-randomized throughout a session and across participants. Control trials were received in either separate blocks between blocked-order experimental trials or randomly interleaved between random-ordered experimental trials. Participants were told when they would be required to track the target with upper limb and eyes, but not whether it would be occluded and what were its motion characteristics. 
Data analysis
Eye position data were differentiated by means of a central difference algorithm to obtain eye velocity and acceleration. Saccades were detected when eye acceleration was above a threshold of 750 deg/s2. When the threshold criteria were exceeded, the peak and trough of acceleration were found such that the complete saccade trajectory could be identified. 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 (30 deg/s) was applied. To obtain desaccaded smooth eye velocity, identified saccades, plus additional 5 data points (equivalent to 25 ms) at the beginning and end of the saccade trajectory, were removed from the eye velocity trace. The removed data were replaced by a linear interpolation routine based on the smooth eye velocity before and after the saccade (for further details, see Bennett & Barnes, 2003). The desaccaded eye velocity data were then low-pass filtered at 25 Hz using an autoregressive zero-phase digital filter. The x-axis arm displacement data sampled by the laser mouse were low-pass filtered using a zero-phase digital filter with a 20 Hz cut-off and then differentiated using a central difference algorithm to obtain velocity. 
In experimental trials, eye and arm velocity at the moment of target disappearance and reappearance (i.e., start and end of occlusion) were extracted from the smoothed data. Minimum eye and arm velocity during occlusion were also calculated but not for data from Experiment 2 because minimum velocity should occur at the moment decelerating targets reappear. To measure participants' ability to match eye and arm displacement to target displacement, total eye displacement (TED) and total arm displacement (TAD) were calculated. For the eye only, the contribution from smooth pursuit and saccades to TED was determined by calculating smooth eye displacement (SED) and saccadic eye displacement (SAD) during occlusion. Intra-participant means for each dependent variable were derived for each combination of independent variables. In order to minimize the influence of a preceding block of trials on a subsequent block, the first trial was excluded from the calculation. For random-order trials, intra-participant means were also calculated using trials two through six. 
For Experiment 1, the dependent variables were submitted to separate 2 condition (ocular, oculo-manual) × 2 order (blocked, random) × 3 velocity (10, 15, or 20 deg/s) × 3 occlusion (600, 800, and 1000 ms) ANOVAs with repeated measures on all factors. For Experiment 2, the data were submitted to separate 2 condition (ocular, oculo-manual) × 2 order (blocked, random) × 5 acceleration (−12, −8, 0, 8 or 12 deg/s2) × 2 occlusion (600, 1000 ms) ANOVAs with repeated measures on all factors. Where appropriate, main and interaction effects were analyzed using Tukey's HSD post-hoc procedure. For the sake of brevity, significant effects only are reported in the text but without p values, which were less than or equal to 0.05. Data from control trials were not included in the analysis but were compared qualitatively. 
Results
Experiment 1
Eye velocity
Participants used the predictable timing of target motion onset to release anticipatory smooth pursuit (see Figure 2), which was then scaled such that eye velocity differed in accord with target velocity at the moment of target disappearance, F(2, 18) = 92.20. There was no difference between random-order and blocked-order trials at this time, irrespective of whether the target was tracked in the ocular or oculo-manual condition. Participants used the available visual feedback to track well the initial target motion with their eyes. 
Figure 2
 
Individual-participants mean (n = 5 trials) smooth eye velocity in ocular (left panels) and oculo-manual (right panels) conditions. Participants were pursuing a 20 deg/s target through a 600 ms occlusion in experimental trials (upper panel) received in blocked order. Control trials (lower panels) in which the 20 deg/s target remained visible throughout are shown for comparison. White (on) and black (off) bars below the abscissa indicate target visibility, which is highlighted by vertical dashed lines that cross the time series data.
Figure 2
 
Individual-participants mean (n = 5 trials) smooth eye velocity in ocular (left panels) and oculo-manual (right panels) conditions. Participants were pursuing a 20 deg/s target through a 600 ms occlusion in experimental trials (upper panel) received in blocked order. Control trials (lower panels) in which the 20 deg/s target remained visible throughout are shown for comparison. White (on) and black (off) bars below the abscissa indicate target visibility, which is highlighted by vertical dashed lines that cross the time series data.
Following the loss of visual feedback at the onset of occlusion, participants continued to track the invisible moving target but were unable to maintain the level of eye velocity (see Figure 2). There was an initial decay in eye velocity irrespective of tracking condition and trial order, which was often followed by a recovery prior to target reappearance. If the initial recovery did not coincide closely with target reappearance in the longer occlusion durations, there was often a further gradual decay or plateau in eye velocity when pursuing the 20 deg/s target. For minimum eye velocity, this pattern of decay resulted in a significant interaction between target velocity and occlusion duration, F(4, 36) = 3.69. Minimum eye velocity was lower when tracking the 20 deg/s target through the 800 and 1000 ms occlusion compared to the 600 ms occlusion (see upper panel Figure 3). Further analysis of minimum eye velocity revealed evidence of scaling to target velocity, F(2, 18) = 92.28, although this was influenced by condition, order and velocity, F(2, 18) = 3.56. When tracking the target in the oculo-manual condition, there was no difference in minimum eye velocity between blocked-order and random-order trials. However, when tracking the target in the ocular condition, minimum eye velocity was higher in blocked-order than in random-order trials for the 20 deg/s target. It would seem, then, that although there was some advantage conveyed by trial order during ocular tracking, this was not sufficient to match the facilitation afforded by the oculo-manual condition when tracking the 20 deg/s target. 
Figure 3
 
Group mean minimum (upper panel) and reappearance (lower panel) eye and arm velocity in experimental trials as a function of condition, trial order, target velocity, and occlusion. Ocular pursuit (Oc) is depicted by solid lines with filled squares, whereas ocular-manual pursuit (OcM) is depicted by solid lines with unfilled squares. Blocked-order and random-order trials are represented by black and grey lines, respectively. Arm velocity is depicted by broken lines with filled circles. Target velocity (TGT) is shown for comparison and is represented by filled red circles.
Figure 3
 
Group mean minimum (upper panel) and reappearance (lower panel) eye and arm velocity in experimental trials as a function of condition, trial order, target velocity, and occlusion. Ocular pursuit (Oc) is depicted by solid lines with filled squares, whereas ocular-manual pursuit (OcM) is depicted by solid lines with unfilled squares. Blocked-order and random-order trials are represented by black and grey lines, respectively. Arm velocity is depicted by broken lines with filled circles. Target velocity (TGT) is shown for comparison and is represented by filled red circles.
As shown in the lower panel of Figure 3, eye velocity was scaled to target velocity at the end of occlusion, F(2, 18) = 81.20, but there was still a large undershoot, with group means of 8.84 ± 1.53, 11.61 ± 2.30, and 12.90 ± 2.92 deg/s for target velocities of 10, 15, and 20 deg/s respectively. The undershoot for each level of target velocity was influenced by occlusion duration, F(4, 36) = 4.77, such that eye velocity was lower when tracking the occluded 20 deg/s target for 1000 ms compared to 600 and 800 ms. There was also a significant interaction between trial order and target velocity, F(2, 18) = 10.23, with greater eye velocity at the end of occlusion evident in blocked order trials than in random order trials when tracking the 20 deg/s target. Finally, as can be seen in the lower panel of Figure 3, there was a significant effect of condition, F(1, 9) = 9.80, with oculo-manual tracking resulting in increased levels of eye velocity at the end of occlusion for each target velocity compared to ocular tracking. 
Eye displacement
Despite the decay and often insufficient recovery in eye velocity in the absence of visual feedback, total eye displacement (TED) was well matched to target displacement (Figure 4) resulting from the different levels of target velocity, F(2, 18) = 1133.75, and occlusion duration, F(2, 18) = 2626.65. Group mean TED differed from target displacement by between −1.2 and 2.2 deg across the different target parameters. There was also a main effect of condition, F(1, 9) = 17.6, which was reflective of greater TED in the ocular compared to oculo-manual condition (Figure 4). Finally, there was a small but significant advantage (∼1.2 deg) in blocked order trials compared to random order trials when tracking the 20 deg/s target through a 1000 ms occlusion, F(4, 36) = 2.89. 
Figure 4
 
Group mean total eye displacement (TED) in experimental trials (collapsed over blocked and random order presentation) as a function of condition, target velocity, and occlusion. Contributions from smooth eye displacement (SED) and saccadic eye displacement (SAD) are represented by filled grey and white bars, respectively. Error bars show the group standard deviation of TED. Target displacement (TGT) is represented by filled red circles.
Figure 4
 
Group mean total eye displacement (TED) in experimental trials (collapsed over blocked and random order presentation) as a function of condition, target velocity, and occlusion. Contributions from smooth eye displacement (SED) and saccadic eye displacement (SAD) are represented by filled grey and white bars, respectively. Error bars show the group standard deviation of TED. Target displacement (TGT) is represented by filled red circles.
Smooth eye displacement (SED) was scaled to the changes in target displacement resulting from the different levels of target velocity, F(2, 18) = 289.38, and occlusion duration, F(2, 18) = 589.53. As can be seen in Figure 4, scaling of SED as a function of increasing velocity was also mediated by condition, F(2, 18) = 7.30. Greater SED was exhibited when tracking 15 and 20 deg/s targets in the oculo-manual condition compared to the ocular condition. In addition, oculo-manual tracking produced greater SED for each occlusion duration compared to ocular tracking alone, F(2, 18) = 6.09. Finally, there was also a small but significant increase in SED in blocked order trials compared to random order trials when tracking the 20 deg/s target, F(2, 18) = 10.38. Still, as expected given the initial decay in smooth pursuit, SED was not perfectly matched to target displacement, and hence there was a contribution from saccadic eye displacement (SAD). While there was no effect of order, there was increased SAD in the ocular condition compared to the oculo-manual condition, F(1, 9) = 21.20. The combination of SAD and SED tended to produce more overshoot of target displacement generated by the 10 deg/s and 15 deg/s targets but a more accurate match for the 20 deg/s target (see Figure 4). 
Arm velocity and displacement
Like the ocular response, participants exhibited anticipatory manual pursuit that was scaled to target velocity 600 ms after the target motion onset, F(2, 18) = 310.81. There was, however, evidence of overshoot in arm velocity compared to target velocity, with group means of 9.77 ± 2.43, 17.62 ± 2.31, and 25.07 deg/s ± 3.30 deg/s for target velocities of 10, 15, and 20 deg/s, respectively. The initial overshoot was most pronounced during pursuit of the 20 deg/s target and was followed by oscillations during the occlusion (Figure 5). The overshoot in arm velocity at the moment of occlusion when tracking the 20 deg/s target was somewhat reduced in blocked-order compared to random-order trials, F(2, 18) = 10.53. 
Figure 5
 
Individual-participants mean (n = 5 trials) arm velocity in blocked-order (left panels) and random-order (right panels) trials. Participants were pursuing a 20 deg/s target through a 600 ms occlusion in experimental (upper panels) trials, or an equivalent duration in control (lower panels) trials. White (on) and black (off) bars below the abscissa indicate target visibility, which is highlighted by vertical dashed lines that cross the time series data. Overall group mean is represented by the thick red line in each panel.
Figure 5
 
Individual-participants mean (n = 5 trials) arm velocity in blocked-order (left panels) and random-order (right panels) trials. Participants were pursuing a 20 deg/s target through a 600 ms occlusion in experimental (upper panels) trials, or an equivalent duration in control (lower panels) trials. White (on) and black (off) bars below the abscissa indicate target visibility, which is highlighted by vertical dashed lines that cross the time series data. Overall group mean is represented by the thick red line in each panel.
During occlusion, minimum arm velocity was scaled to different target velocities, F(2, 18) = 217.77. This scaling was mediated by occlusion duration, F(4, 36) = 6.73, with minimum arm velocity when tracking 10 deg/s and 15 deg/s targets being equal across each occlusion, and reduced as occlusion duration increased when tracking the 20 deg/s target (see upper panel Figure 3). At the moment of target reappearance, arm velocity remained scaled to target velocity, F(2, 18) = 144.73, and resulted in group means of 9.50 ± 0.63, 14.96 ± 1.15 and 20.51 ± 2.02 deg/s when tracking target velocities of 10, 15, and 20 deg/s respectively. Moreover, while arm velocity was better maintained during occlusion than was eye velocity, there was an interaction with occlusion duration, F(4, 36) = 8.64, which was a result of overshoot and undershoot in hand velocity when tracking the 20 deg/s target over a 600 and 1000 ms occlusion, respectively (see lower panel Figure 3). The arm velocity trajectory resulted in total arm displacement (TAD) during occlusion that was well matched to target displacement resulting from different levels of target velocity, F(2, 18) = 535.97, and occlusion duration, F(2, 18) = 264.21. Overall, the error in matching arm displacement to target displacement ranged between −0.3 deg and 1.8 deg across the different target parameters, and was comparable in accuracy to eye displacement. Finally, it is notable that there were no main or interaction effects involving trial order. 
Experiment 2
Eye velocity
Having used the predictable timing of target motion onset to release anticipatory smooth pursuit (left panel Figure 6), participants continued to track the moving target, albeit with eye velocity that was not ideally scaled to target velocity. This scaling was reflected in a main effect of acceleration for eye velocity at the moment of target disappearance, F(4, 44) = 83.85. Importantly, though, whereas participants did initially respond by moving their eyes in accord with the different target velocities at onset (i.e., 19.2, 16.8, 12, 7.2, and 4.8 deg/s), they subsequently modified smooth eye velocity in accord with target acceleration. Eye velocity at target disappearance was also subject to an interaction between condition and acceleration, F(4,44) = 11.06. Oculo-manual pursuit resulted in increased eye velocity with negatively accelerating targets compared to ocular pursuit. Finally, there was an interaction between order and acceleration, F(4, 44) = 22.34. Eye velocity at target disappearance was greater in random-order than in blocked-order trials for negatively accelerating targets. Notably, this difference was reflected in a linear decrease in eye velocity in random-order trials, which is consistent with somewhat poorer scaling to target acceleration. 
Figure 6
 
Individual participant (P5) mean (n = 5 trials) smooth eye (left panel) and arm (right panel) velocity in oculo-manual conditions. Blocked-order and random-order experimental trials are represented as solid black and grey lines, respectively. Participant was pursuing targets accelerating at −12, 0, or 12 deg/s2 through a 1000 ms occlusion. White (on) and black (off) bars below the abscissa indicate target visibility, which is highlighted by vertical dashed lines that cross the time series data. Red solid lines represent the mean smooth eye (left panel) and arm (right panel) velocity in blocked-order control trials.
Figure 6
 
Individual participant (P5) mean (n = 5 trials) smooth eye (left panel) and arm (right panel) velocity in oculo-manual conditions. Blocked-order and random-order experimental trials are represented as solid black and grey lines, respectively. Participant was pursuing targets accelerating at −12, 0, or 12 deg/s2 through a 1000 ms occlusion. White (on) and black (off) bars below the abscissa indicate target visibility, which is highlighted by vertical dashed lines that cross the time series data. Red solid lines represent the mean smooth eye (left panel) and arm (right panel) velocity in blocked-order control trials.
As shown in Figure 6 (left panel), participants continued to track the moving target during occlusion, and in trials with negative acceleration there was a continuous decrease in eye velocity that resulted in a good match to target velocity at the moment of reappearance. For constant velocity and positively accelerating targets, there was an initial decay in eye velocity irrespective of condition and trial order, which was often followed by an anticipatory recovery. The recovery was, however, insufficient to return eye velocity to the level of target velocity at the moment of reappearance, particularly when pursuing positively accelerating targets (i.e., 8 and 12 deg/s2). The anticipatory recovery was subject to an interaction between trial order, acceleration and occlusion F(4, 44) = 4.96; p < 0.05. As shown in Figure 7, eye velocity at reappearance was better scaled during blocked-order than during random-order trials when tracking constant velocity and positively accelerating targets for both a 600 and 1000 ms occlusion. The same effect was present when tracking negatively accelerating targets for a 1000 ms occlusion. Contrary to Experiment 1, there was no effect of condition on eye velocity at the moment of target reappearance. Therefore, it would seem that when presented with constant velocity and accelerating targets, increased predictability of target trajectory during blocked-order compared to random-order trials exerted a more powerful effect on eye velocity during occlusion than did the availability of sensorimotor signals in oculo-manual pursuit. 
Figure 7
 
Group mean reappearance eye and arm velocity in experimental trials as a function of condition, trial order, target acceleration, and occlusion (600 ms – upper panel; 1000 ms - lower panel). Ocular pursuit (Oc) is depicted by solid lines with filled squares, whereas ocular-manual pursuit (OcM) is depicted by solid lines with unfilled squares. Blocked-order and random-order trials are represented by black and grey lines, respectively. Arm velocity is depicted by broken lines with filled circles and follows the color coding for trial order. Target velocity (TGT) is shown for comparison and is represented by filled red circles.
Figure 7
 
Group mean reappearance eye and arm velocity in experimental trials as a function of condition, trial order, target acceleration, and occlusion (600 ms – upper panel; 1000 ms - lower panel). Ocular pursuit (Oc) is depicted by solid lines with filled squares, whereas ocular-manual pursuit (OcM) is depicted by solid lines with unfilled squares. Blocked-order and random-order trials are represented by black and grey lines, respectively. Arm velocity is depicted by broken lines with filled circles and follows the color coding for trial order. Target velocity (TGT) is shown for comparison and is represented by filled red circles.
Eye displacement
With the exception of negatively accelerating and constant velocity targets, total eye displacement (TED) during occlusion undershot target displacement (Figure 8). Tracking condition modified control of eye displacement, as reflected by an interaction between condition and ISI; F(1, 11) = 6.23. There was a small significant increase (0.5 deg) in TED during ocular compared to oculo-manual pursuit over the 1000 ms occlusion. A more notable difference in TED was evident in the interaction between order, acceleration and ISI, F(4, 44) = 18.51. TED over a 1000 ms occlusion in random-order trials overshot displacement of negatively accelerating targets (−12 deg/s2 = 2.42 deg, −8 deg/s2 = 1.04 deg) and undershot displacement of positively accelerating targets (8 deg/s2 = −4.70 deg, 12 deg/s2 = −6.34 deg). In blocked-order trials, there was much smaller undershoot in TED for positively accelerating targets over the 1000 ms occlusion (8 deg/s2 = −2.03 deg, 12 deg/s2 = −2.21 deg). There was almost no overshoot of target displacement of negatively accelerating targets in blocked-order trials. 
Figure 8
 
Group mean total eye displacement (TED) in experimental trials as a function of trial order, target acceleration, and occlusion. Contributions from smooth eye displacement (SED) and saccadic eye displacement (SAD) are represented by filled grey and white bars, respectively. Error bars show the group standard deviation of TED. Target displacement (TGT) is represented by filled red circles
Figure 8
 
Group mean total eye displacement (TED) in experimental trials as a function of trial order, target acceleration, and occlusion. Contributions from smooth eye displacement (SED) and saccadic eye displacement (SAD) are represented by filled grey and white bars, respectively. Error bars show the group standard deviation of TED. Target displacement (TGT) is represented by filled red circles
Separating TED into its individual components also revealed an interaction between order, acceleration and ISI for both SED, F(4, 44) = 7.13, and SAD, F(4, 44) = 7.94. Over a 1000 ms occlusion, SED was greater in blocked-order than in random-order trials when tracking target that accelerated at −12, 8, and 12 deg/s2. Similarly, SAD was increased in blocked-order compared to random-order trials when tracking targets that accelerated at 8 and 12 deg/s2 targets.As can be seen in Figure 8, the co-variation between SED and SAD, and hence matching of TED to target displacement, was more effective when trials were received in blocked order. 
Arm velocity and displacement
As with the ocular response, participants exhibited anticipatory manual pursuit. Also, they experienced difficulty scaling arm velocity to target velocity at the moment of occlusion (right panel Figure 6). There was a large overshoot when tracking negatively accelerating targets, and an undershoot when tracking positively accelerating targets, resulting in a main effect of target acceleration, F(4, 44) = 87.98. Hand velocity at this moment was also influenced by an interaction between order and acceleration, F(4, 44) = 7.37. Compared to random-order trials, blocked-order trials resulted in a better match between hand velocity and target velocity at the moment of occlusion, and particularly when tracking −12 and 12 deg/s2 targets. 
The pattern of overshoot and undershoot in arm velocity at disappearance largely remained throughout occlusion, culminating at the moment of target reappearance in an interaction between order, acceleration and ISI, F(4, 44) = 6.38. Arm velocity followed a similar linear increase as a function of target acceleration for both random-order and blocked trials over a 600 ms occlusion. However, over a 1000 ms occlusion, arm velocity in blocked-order trials was better matched to target velocity of both −12 deg/s2 and 12 deg/s2 targets than in comparable random-order trials (Figure 7). As expected, this pattern of results was reflected in total hand displacement (THD) during occlusion, which was also subject to an interaction between order, acceleration and ISI, F(4, 44) = 11.86. There was a better match of THD to target displacement after 1000 ms of tracking the −12 and 12 deg/s2 targets in blocked-order than random-order trials. 
Discussion
Contrary to smooth ocular pursuit, control of distal effectors such as the arm is not so dependent on visual feedback. Instead, proprioception (afference) from muscles and joints can be used to confirm the output of internal drive (e.g., efference copy), thus enabling arm movement to continue at will during occlusion. The availability of these sensorimotor signals from the upper limb can also be of benefit in maintaining smooth pursuit eye movements of self-generated target motion in darkness. For instance, deafferentation (Gauthier & Ivaldi, 1988) or ischaemic block (Gauthier & Hofferer, 1976) results in almost no smooth pursuit of an imagined target attached to the participant's finger, thus indicating that limb afference makes a necessary and sufficient contribution. Similarly, when instructed to stop moving the limb to which an imagined target is attached, smooth pursuit quickly decays and is replaced by saccades that locate the eyes reasonably well relative to the unseen target (Gauthier & Hofferer, 1976). Such an inability to maintain smooth pursuit in the absence of visual and proprioceptive feedback is somewhat at odds with work on pursuit of externally-generated target motion (Becker & Fuchs, 1985; Madelain & Krauzlis, 2003; Pola & Wyatt, 1997). Indeed, despite exhibiting a rapid decay in smooth pursuit following target occlusion, participants often exhibit an anticipatory, and sometimes predictive, recovery when they expect the target to reappear (Bennett & Barnes, 2003; Bennett & Barnes, 2004; Orban de Xivry et al., 2006). 
The current study examined whether the availability of sensorimotor signals from concurrent arm tracking of externally-generated target motion affords a similar advantage to ocular pursuit as that when tracking self-generated target motion, and how this advantage is manifest in eye velocity and displacement. Importantly, because externally-generated target motion is not the product of sensorimotor signals, there is opportunity for greater independence between the arm and ocular response. Consequently, difficulties observed previously when switching (within-trial) between oculo-manual and ocular pursuit (Gauthier & Hofferer, 1976) are less likely to impact eye movements. In addition, by using externally generated target motion, it is possible to examine how predictability (i.e., trials arranged in random order or blocked order) influences the ocular response during transient occlusion, and also whether there is mediation by the contribution from sensorimotor input. 
Not surprisingly, in both experiments reported here, we observed that the ocular response was not as well maintained during transient occlusion as it was in control trials where visual feedback was available throughout (Becker & Fuchs, 1985). We also confirmed that ocular pursuit during occlusion can be improved by arranging trials in blocked rather than in random order (Bennett & Barnes, 2003; Bennett et al., 2010). As expected, this was not the case for all combinations of target parameters. For eye velocity at reappearance, facilitation was most evident in Experiment 1 for the fastest moving target (20 deg/s) over the longest occlusion duration (1000 ms), and in experiment 2 for the outermost negatively and positively accelerating targets (−12 and 12 deg/s2). 
This result was reflected in smooth eye displacement, and hence total eye displacement, which were both more accurate when pursuing −12 and 12 deg/s2 targets over the longest occlusion during blocked-order than random-order trials. As a consequence of this better correspondence between smooth eye displacement and target displacement for negatively accelerating targets, trial order did not influence saccadic eye displacement. The effect of target predictability was also more prevalent on the manual response when pursuing accelerating targets. For instance, in Experiment 2, arm velocity at the end of the longer occlusion was better matched to target velocity for the most negative and positive levels of acceleration in blocked-order trials than in comparable random-order trials. This effect was reflected in total hand displacement during occlusion, where there was a better match to target displacement. 
A somewhat different pattern of results between the two experiments was found for the effect of tracking condition. In Experiment 1, oculo-manual tracking resulted in increased eye velocity at the end of occlusion for each target velocity compared to ocular tracking. This effect was reflected in increased smooth eye displacement, and thus a better match between total eye displacement and target displacement for the majority of combinations of target velocity and occlusion duration. Notably, the effect of producing concurrent arm movement on ocular control was greater than that of trial order. In Experiment 2, however, there was no meaningful difference in eye velocity at the end of occlusion, or measures of eye displacement, between the oculo-manual and ocular conditions. The lack of difference was not because the ocular response was better maintained during occlusion when tracking positively accelerating compared to constant velocity targets. Eye velocity and displacement error were of considerably greater magnitude when tracking the 12 deg/s2 in Experiment 2 (see Figures 7 and 8) compared to the 20 deg/s target in Experiment 1 (see Figures 4 and 5). The increased error following pursuit of positively accelerating targets also cannot simply be explained by the magnitude of target velocity at reappearance and displacement during occlusion.Target velocity at reappearance with 8 and 12 deg/s2 acceleration was 20 and 24 deg/s, respectively, whereas target displacement during occlusion was 16 and 18 deg/s, which is comparable to those of the fastest target over the longest occlusion in Experiment 1 (i.e., 20 deg/s and 20 deg). Finally, it is clear that the lack of facilitation to the ocular response during oculo-manual pursuit of positively accelerating targets was not a result of a limitation in control of arm movement. Arm velocity and displacement were better maintained during occlusion, resulting in considerably less undershoot than was evident for the eye. It would seem, then, that pursuing the occluded trajectory of positively accelerating targets with the eyes remains problematic even when trials are received in blocked order (Bennett et al., 2010), and, further, that having access to sensorimotor signals from concurrent arm movement does not facilitate prediction in the ocular response. 
It is relevant to comment here that the sensorimotor transformations required in the current study were quite different from those in the majority of studies on smooth pursuit eye movements of self-generated target motion (Gauthier & Ivaldi, 1988; Gauthier & Hofferer, 1976). In particular, arm motion was constrained on a linear rail parallel to the screen, with a non-unity gain relationship between arm and target displacement. Participants adapted to the required sensorimotor transformation between arm motion and its visual consequences (see below for more detail), but still it is possible that under such circumstances the ocular control system was unable to fully take advantage. In this respect, it is notable that Mather and Lackner (1981) reported facilitation of ocular pursuit under normal illumination or in darkness when the visible or imagined target was the result of spatially congruent (i.e., flexion/extension of forearm) or incongruent (elbow extension/flexion and shoulder flexion/extension) sinusoidal movement. Findings of qualitatively similar smooth pursuit when tracking self-generated spatially congruent or incongruent visible target motion were also reported by Steinbach (1969). In these studies, though, self-generated target motion was cyclical, lasting 10 to 30 s, and did not undergo discrete occlusion (i.e., lights on or off for the duration of testing). Also, relatively gross measures of ocular pursuit were reported (e.g., saccades per cycle, number of saccades, eye-arm amplitude gain, percentage of time spent in smooth pursuit), thus making it difficult to compare the magnitude of facilitation by concurrent arm movement. It will be interesting in future work to determine if, and how, facilitation develops with more extensive experience of externally-generated target motion. For instance, it is possible that there could be greater facilitation of ocular pursuit during discrete occlusions introduced in each half-cycle of sinusoidal target movement (i.e., non-constant velocity). Indeed, it is known that more extensive practice results in improved predictive (feedforward) control of ocular (Madelain & Krauzlis, 2003) and manual pursuit (Miall & Jackson, 2006), and that oculo-manual coordination in children quickly improves to the levels of adults, with the eyes and the hand seeming to exhibit independence (see Gauthier et al., 1988). 
As described above, findings for arm movement were somewhat different from those of the eyes. Unlike the ocular response, there was a large overshoot in arm velocity compared to target velocity at the moment of occlusion (see also control trials for the same effect), except when pursuing positively accelerating targets in Experiment 2. Similar effects were reported by Barnes and Marsden (2002), who suggested that the overshoot was reflective of positional control for arm movement having different dynamics (e.g., lower peak velocity) than that of saccadic eye movements. In experimental trials, this initial overshoot in arm velocity was followed by a reduction and then continuation (constant velocity and positively accelerating targets) that was maintained closer to target velocity than was achieved by the eye. For negatively accelerating targets, there was a continual reduction in arm velocity that matched well target velocity. Accordingly, arm velocity and displacement was better matched to the occluded target trajectory than that of the eye. In control trials where the target remained visible, there was evidence of greater oscillation in arm velocity than in smooth eye velocity. Again, this difference can be explained by the different dynamics of intermittent error control mechanisms in the ocular and manual systems. Also, for constant velocity targets there was greater oscillation in arm velocity during control trials than in experimental trials. Presumably, control of the arm during occlusion was based on a comparison of sensorimotor signals (i.e., limb afference and efference) to an internal representation of target motion, and hence was not biased by visually-based corrections (Miall, Weir, & Stein, 1993). 
Arranging the trials in blocked as opposed to random order resulted in some evidence of better correspondence between the arm and target in experimental trials. There was less overshoot in arm velocity at the moment of occlusion when tracking the 20 deg/s target in blocked-order trials (Experiment 1), and a better match between arm velocity and target velocity at the moment of occlusion when tracking −12 and 12 deg/s2 targets (Experiment 2). However, the effect of trial order during occlusion was only evident in Experiment 2, where arm velocity at the end of a 1000 ms occlusion in blocked-order trials was better matched to target velocity of both −12 and 12deg/s2 targets than in comparable random-order trials. As expected, this pattern of results was reflected in total hand displacement during occlusion. These findings show that participants benefited during occlusion in experimental trials from the opportunity to represent and consolidate target acceleration over repeated attempts, as well as the availability of long-term (i.e., between trial) prediction in the form of implicit advance knowledge (Bennett et al., 2010). This was not evident in control trials where participants could use the available visual feedback to control arm motion. Overall, then, these findings show participants were quickly able to adapt to the required sensorimotor transformations and achieve good coupling. This was no doubt aided by the inclusion of familiarization trials prior to the experiment, as well as experience of using peripheral devices to interact with computer displays during normal daily life. 
Eye and arm coupling
While the design and analysis of the current study was focused on the issue of facilitation of ocular pursuit by sensorimotor signals, the findings from discrete measures of position and velocity control are qualitatively consistent with the suggestion of independent coupling between the ocular and manual systems in a situation where they are attempting to achieve the same goal (Bock, 1987; Gauthier et al., 1988; Lazzari et al., 1997; Scarchilli & Vercher, 1999; Vercher et al., 1997). Such oculo-manual interaction has been modeled (e.g., coordination control system) with a convergence of signals from the eyes and upper limb in the cerebellum, which is known to exhibit increased activity when performing coordinated oculo-manual tracking (Miall, Reckess, & Imamizu, 2001), as well as being associated with impaired motor control in individuals with cerebellar lesions (van Donkelaar & Lee, 1994; Vercher & Gauthier, 1988). According to this scheme, facilitation of ongoing smooth pursuit by concurrent upper limb movement is said to be the result of modifying a gain function (i.e., mutual coupling), the magnitude of which is dependent on the (auto and cross) correlation between sensorimotor signals and target motion. The gain function is suggested to act on an internal positive feedback loop within the smooth pursuit branch, which is the basis of short-term prediction in smooth pursuit to ongoing target motion (Bahill & McDonald, 1983; Robinson, Gordon, & Gordon, 1986). Limb afference is suggested to be both necessary and sufficient (Gauthier & Hofferer, 1976; Gauthier & Ivaldi, 1988), but only increases the internal drive to ocular pursuit if upper limb movement is closely related to externally-generated motion, or the cause of target motion (i.e., self-generated). 
There are similarities with this model of oculo-manual control and a model we formulated to account for the anticipatory and predictive recovery of smooth pursuit during head-fixed (Barnes, 2008; Bennett & Barnes, 2004; Bennett et al., 2007) or head-free (Ackerley & Barnes, 2011) transient occlusion (Figure 9). Similar to traditional accounts, the model comprises a visual feedback loop and an internal efference copy loop. Recent evidence indicates that the efference copy is sampled and held as a reference level signal that is then fed through either a direct loop during randomized trials or an indirect loop when stimuli are blocked and thus more predictable. The common output of direct and indirect loops has a gain β (< = 1) that is dependent on expectation (e.g., target reappearance) and conflict detection. It is assumed that target occlusion, by removing retinal error information, creates conflict between predicted and actual target velocity and which causes a temporary reduction in β and consequent reduction in smooth eye velocity. However, expectation of imminent target reappearance enables β to return to its pre-occlusion level, thereby generating an anticipatory increase in eye velocity back towards the reference level. The indirect loop includes a short-term store (MEM) that accumulates prior efference copy samples and builds up a more complete representation of target motion when stimuli are blocked (i.e., predictive), thus allowing higher levels of eye velocity during occlusion. 
Figure 9
 
Model of ocular pursuit receiving retinal and extra-retinal inputs. Retinal input is arranged as a negative visual feedback pathway. Extra-retinal input is received from either a direct or indirect (predictive) loop. When there is no expectation of future target movement, sw2 is set to allow the direct loop to operate, whereas when expectation is high (i.e., motion is predictable), sw2 is set to allow positive feedback via the short term memory (MEM) which holds a reference level of velocity captured from previous samples and can be used to predict target motion. The input to both direct and indirect pathways comes from sampling and holding a copy of the visuomotor drive signal (vmd) in module S/H. It is assumed that the gain β is modified by expectation (max value unity) and by conflict (or mismatch) detection. Sudden withdrawal of input causes a temporary reduction of gain β, so that eye velocity initially falls, but can then recover (e.g., towards the end of a transient occlusion). The inclusion of a more robust short-term store, MEM, within the indirect loop includes enables velocity information to be held over longer periods and during fixation. The gain, α(t), at the output of MEM can also be modulated as a function of time to allow more complex motion trajectories to be produced in anticipation of expected target motion (e.g., negative or positive acceleration) visual feedback delay – τv = 0.1 s.
Figure 9
 
Model of ocular pursuit receiving retinal and extra-retinal inputs. Retinal input is arranged as a negative visual feedback pathway. Extra-retinal input is received from either a direct or indirect (predictive) loop. When there is no expectation of future target movement, sw2 is set to allow the direct loop to operate, whereas when expectation is high (i.e., motion is predictable), sw2 is set to allow positive feedback via the short term memory (MEM) which holds a reference level of velocity captured from previous samples and can be used to predict target motion. The input to both direct and indirect pathways comes from sampling and holding a copy of the visuomotor drive signal (vmd) in module S/H. It is assumed that the gain β is modified by expectation (max value unity) and by conflict (or mismatch) detection. Sudden withdrawal of input causes a temporary reduction of gain β, so that eye velocity initially falls, but can then recover (e.g., towards the end of a transient occlusion). The inclusion of a more robust short-term store, MEM, within the indirect loop includes enables velocity information to be held over longer periods and during fixation. The gain, α(t), at the output of MEM can also be modulated as a function of time to allow more complex motion trajectories to be produced in anticipation of expected target motion (e.g., negative or positive acceleration) visual feedback delay – τv = 0.1 s.
Facilitation of smooth pursuit by concurrent arm movement in the above model can be explained by the presence of sensorimotor signals influencing the expectation that the target will continue to move and reappear later in its trajectory (cf. Gauthier & Hofferer, 1976). Indeed, confirmation of the output from feedforward motor drive by sensory feedback is a cornerstone of predictive motor control (Barnes & Marsden, 2002). In addition, the finding that concurrent arm movement facilitated ocular pursuit in both random-order and blocked-order trials, is consistent with the placement of β such that it acts independently on the direct and indirect loops. Importantly, though, because MEM is located within the indirect loop, where it enables better representation of occluded target motion and thereby a predictive recovery scaled to post-occlusion target velocity, facilitation of ocular pursuit by trial order does not necessarily interact with concurrent2 arm movement. This is exactly what we found in the current study. Finally, it is also notable that a step change in β when performing concurrent arm movement will result in facilitation of smooth pursuit that is somewhat independent of the ongoing arm movement. This feature of the model can explain the different ability to maintain ocular and manual pursuit during experimental and control trials, as well as at target motion onset (see also Barnes & Marsden, 2002). Notably, this relative independence has also recently been observed in the control of head and eye movement during head-free pursuit (Ackerley & Barnes, 2011). 
Conclusion
We have shown here that concurrent arm movement and/or increased predictability regarding the upcoming target motion can facilitate smooth ocular pursuit during a transient occlusion. However, for some characteristics of externally-generated target motion, there is an upper limit on volitional control of smooth pursuit during occlusion. The same is not true for arm movement, which matches well target motion in the absence of visual feedback. These findings show that while predictability and/or sensorimotor signals influence the ocular response during transient occlusion, they are not able to completely substitute for the lack of visual feedback. 
Acknowledgments
The research was supported in part by a grant from the Medical Research Council, UK, to G. R. Barnes, and a Natural Sciences and Engineering Research Council of Canada (NSERC) Post-Doctoral award (Canada-United Kingdom Millennium Fellowship Program) to S. Hansen. 
Commercial relationships: none. 
Corresponding author: Simon J Bennett. 
Email: S.J.Bennett@ljmu.ac.uk. 
Address: Research Institute for Exercise and Sport Sciences, Liverpool John Moores University, Liverpool, UK. 
References
Ackerley R. Barnes G. R. (2011). The interaction of visual, vestibular and extra-retinal mechanisms in the control of head and gaze during head-free pursuit. Journal of Physiology, 589, 1627– 1642. [CrossRef] [PubMed]
Bahill A. T. McDonald J. D. (1983). Model emulates human smooth pursuit system producing zero-latency target tracking. Biological Cybernetics, 48, 213– 222. [CrossRef] [PubMed]
Barnes G. R. (2008). Cognitive processes involved in smooth pursuit eye movements. Brain and Cognition, 68, 309– 326. [CrossRef] [PubMed]
Barnes G. R. Marsden J. F. (2002). Anticipatory control of hand and eye movements in humans during oculo-manual tracking. Journal of Physiology, 539, 317– 330. [CrossRef] [PubMed]
Becker W. Fuchs A. F. (1985). Prediction in the oculomotor system: Smooth pursuit during transient disappearance of a visual target. Experimental Brain Research, 57, 562– 575. [CrossRef] [PubMed]
Bennett S. J. Barnes G. R. (2003). Human ocular pursuit during the transient disappearance of a moving target. Journal of Neurophysiology, 90, 2504– 2520. [CrossRef] [PubMed]
Bennett S. J. Barnes G. R. (2004). Predictive smooth ocular pursuit during the transient disappearance of a visual target. Journal of Neurophysiology, 92, 578– 590. [CrossRef] [PubMed]
Bennett S. J. Barnes G. R. (2005). Timing the anticipatory recovery in smooth ocular pursuit during the transient disappearance of a visual target. Experimental Brain Research, 163, 198– 203. [CrossRef] [PubMed]
Bennett S. J. Orban de Xivry J.-J. Barnes G. R. Lefèvre P. (2007). Target acceleration can be extracted and represented within the predictive drive to ocular pursuit. Journal of Neurophysiology, 98, 1405– 1414. [CrossRef] [PubMed]
Bennett S. J. Orban de Xivry J.-J. Lefèvre P. Barnes G. R. (2010). Oculomotor prediction of accelerative target motion during occlusion: long-term and short-term effects. Experimental Brain Research, 204, 493– 504. [CrossRef] [PubMed]
Bock O. (1987). Coordination of arm and eye movements in tracking of sinusoidally moving targets. Behavioural Brain Research, 24, 93– 100. [CrossRef] [PubMed]
Brouwer A. M. Brenner E. Smeets J. B. J. (2002). Perception of acceleration with short presentation time: Can acceleration be used in interception? Perception & Psychophysics, 64, 1160– 1168. [CrossRef] [PubMed]
Deno D. C. Crandall W. F. Sherman K. Keller E. L. (1995). Characterization of prediction in the primate visual smooth pursuit system. Biological Systems, 34, 107– 128.
Gauthier G. M. Hofferer J. M. (1976). Eye tracking of self-moved targets in the absence of vision. Experimental Brain Research, 26, 121– 139. [CrossRef] [PubMed]
Gauthier G. M. Ivaldi F. M. (1988). Ocular-manual tracking of visual targets in monkey: role of the arm afferent information in the control of arm and eye movements. Experimental Brain Research, 73, 138– 154. [CrossRef] [PubMed]
Gauthier G. M. Vercher J.-L. Ivaldi F. M. Marchetti E. (1988). Ocular-manual tracking of visual targets: control learning, coordination control and coordination model. Experimental Brain Research, 73, 127– 137. [CrossRef] [PubMed]
Lazzari S. Vercher J.-L. Buizza A. (1997). Manuo-ocular coordination in target tracking. I. A model simulating human performance. Biological Cyberbetics, 77, 1– 10. [CrossRef]
Madelain L. Krauzlis R. J. (2003). Effects of learning on smooth pursuit during transient disappearance of a visual target. Journal of Neurophysiology, 90, 972– 982. [CrossRef] [PubMed]
Mather J. A. Lackner J. R. (1981). The influence of efferent, proprioceptive, and timing tactors on the accuracy of eye-hand tracking. Experimental Brain Research, 43, 406– 412. [PubMed]
Miall R. C. Jackson J. K. (2006). Adaptation to visual feedback delays in manual tracking: evidence against the Smith Predictor model of human visually guided action. Experimental Brain Research, 172, 77– 84. [CrossRef] [PubMed]
Miall R. C. Reckess G. Z. Imamizu H. (2001). The cerebellum coordinates eye and hand tracking movements. Nature Neuroscience, 4, 638– 644. [CrossRef] [PubMed]
Miall R. C. Weir D. J. Stein J. F. (1993). Intermittency in human manual tracking tasks. Journal of Motor Behavior, 25, 53– 63. [CrossRef] [PubMed]
Mitrani L. Dimitrov G. (1978). Pursuit eye movements of a disappearing moving target. Vision Research, 18, 537– 539. [CrossRef] [PubMed]
Orban de Xivry J.-J. Bennett S. J. Lefèvre P. Barnes G. R. (2006). Evidence for synergy between saccades and smooth pursuit during transient target disappearance. Journal of Neurophysiology, 95, 418– 427. [CrossRef] [PubMed]
Pola J. Wyatt H. J. (1997). Offset dynamics of human smooth pursuit eye movements: effects of target presence and subject attention. Vision Research, 39, 2767– 2775.
Robinson D. A. Gordon J. L. Gordon S. E. (1986). A Model of the smooth pursuit eye movement system. Biological Cybernetics, 55, 43– 57. [CrossRef] [PubMed]
Scarchilli K. Vercher J.-L. (1999). The oculomanual coordination control center takes into account the mechanical properties of the arm. Experimental Brain Research, 124, 42– 52. [CrossRef] [PubMed]
Steinbach M. J. (1969). Eye tracking of self-moved targets: the role of efference. Journal of Experimental Psychology, 82, 366– 376. [CrossRef] [PubMed]
van Donkelaar P. Lee R. G. (1994). Interactions between the eye and hand motor systems: Disruptions due to cerebellar dysfunction. Journal of Neurophysiology, 72, 1674– 1685. [PubMed]
Vercher J.-L. Gauthier G. M. (1988). Cerebellar involvement in the coordination control of the oculo-manual tracking system: Effects of cerebellar dentate nucleus lesion. Experimental Brain Research, 73, 155– 166. [CrossRef] [PubMed]
Vercher J.-L. Gauthier G. M. Guedon O. Blouin J. Cole J. Lamarre Y. (1996). Self-moved target eye tracking in control and deafferented subjects: Roles of arm motor command and proprioception in arm-eye coordination. Journal of Neurophysiology, 76, 1133– 1144. [PubMed]
Vercher J.-L. Lazzari S. Gauthier G. M. (1997). Manuo-ocular coordination in target tracking. II. Comparing the model with human behavior. Biological Cybernetics, 77, 267– 275. [CrossRef] [PubMed]
Vercher J.-L. Quaccia D. Gauthier G. M. (1995). Oculo-manual coordination control: Respective role of visual and non-visual information in ocular tracking of self-moved targets. Experimental Brain Research, 103, 311– 322. [CrossRef] [PubMed]
Figure 1
 
Schematic diagram of an experimental trial in the ocular (panel A) and ocular-manual (panel B) conditions. A stationary green target (box 1) was presented for 2600 ms on a black screen before changing color to red (box 2). In ocular-manual conditions, an unfilled white circle represented the motion of the hand-held mouse. After a random period between 1650 and 1850 ms, the red target then disappeared for 300 ms (box 3) and reappeared moving horizontally to the right for 600 ms (box 4). The target was then occluded for 600, 800, or 1000 ms (box 5). The target reappeared moving to the right for 400 ms (box 6), and was then extinguished leaving a blank screen (box 7) before the start of another trial. White arrows represent direction when the target was in motion and were not visible to the subject at any part of the trial.
Figure 1
 
Schematic diagram of an experimental trial in the ocular (panel A) and ocular-manual (panel B) conditions. A stationary green target (box 1) was presented for 2600 ms on a black screen before changing color to red (box 2). In ocular-manual conditions, an unfilled white circle represented the motion of the hand-held mouse. After a random period between 1650 and 1850 ms, the red target then disappeared for 300 ms (box 3) and reappeared moving horizontally to the right for 600 ms (box 4). The target was then occluded for 600, 800, or 1000 ms (box 5). The target reappeared moving to the right for 400 ms (box 6), and was then extinguished leaving a blank screen (box 7) before the start of another trial. White arrows represent direction when the target was in motion and were not visible to the subject at any part of the trial.
Figure 2
 
Individual-participants mean (n = 5 trials) smooth eye velocity in ocular (left panels) and oculo-manual (right panels) conditions. Participants were pursuing a 20 deg/s target through a 600 ms occlusion in experimental trials (upper panel) received in blocked order. Control trials (lower panels) in which the 20 deg/s target remained visible throughout are shown for comparison. White (on) and black (off) bars below the abscissa indicate target visibility, which is highlighted by vertical dashed lines that cross the time series data.
Figure 2
 
Individual-participants mean (n = 5 trials) smooth eye velocity in ocular (left panels) and oculo-manual (right panels) conditions. Participants were pursuing a 20 deg/s target through a 600 ms occlusion in experimental trials (upper panel) received in blocked order. Control trials (lower panels) in which the 20 deg/s target remained visible throughout are shown for comparison. White (on) and black (off) bars below the abscissa indicate target visibility, which is highlighted by vertical dashed lines that cross the time series data.
Figure 3
 
Group mean minimum (upper panel) and reappearance (lower panel) eye and arm velocity in experimental trials as a function of condition, trial order, target velocity, and occlusion. Ocular pursuit (Oc) is depicted by solid lines with filled squares, whereas ocular-manual pursuit (OcM) is depicted by solid lines with unfilled squares. Blocked-order and random-order trials are represented by black and grey lines, respectively. Arm velocity is depicted by broken lines with filled circles. Target velocity (TGT) is shown for comparison and is represented by filled red circles.
Figure 3
 
Group mean minimum (upper panel) and reappearance (lower panel) eye and arm velocity in experimental trials as a function of condition, trial order, target velocity, and occlusion. Ocular pursuit (Oc) is depicted by solid lines with filled squares, whereas ocular-manual pursuit (OcM) is depicted by solid lines with unfilled squares. Blocked-order and random-order trials are represented by black and grey lines, respectively. Arm velocity is depicted by broken lines with filled circles. Target velocity (TGT) is shown for comparison and is represented by filled red circles.
Figure 4
 
Group mean total eye displacement (TED) in experimental trials (collapsed over blocked and random order presentation) as a function of condition, target velocity, and occlusion. Contributions from smooth eye displacement (SED) and saccadic eye displacement (SAD) are represented by filled grey and white bars, respectively. Error bars show the group standard deviation of TED. Target displacement (TGT) is represented by filled red circles.
Figure 4
 
Group mean total eye displacement (TED) in experimental trials (collapsed over blocked and random order presentation) as a function of condition, target velocity, and occlusion. Contributions from smooth eye displacement (SED) and saccadic eye displacement (SAD) are represented by filled grey and white bars, respectively. Error bars show the group standard deviation of TED. Target displacement (TGT) is represented by filled red circles.
Figure 5
 
Individual-participants mean (n = 5 trials) arm velocity in blocked-order (left panels) and random-order (right panels) trials. Participants were pursuing a 20 deg/s target through a 600 ms occlusion in experimental (upper panels) trials, or an equivalent duration in control (lower panels) trials. White (on) and black (off) bars below the abscissa indicate target visibility, which is highlighted by vertical dashed lines that cross the time series data. Overall group mean is represented by the thick red line in each panel.
Figure 5
 
Individual-participants mean (n = 5 trials) arm velocity in blocked-order (left panels) and random-order (right panels) trials. Participants were pursuing a 20 deg/s target through a 600 ms occlusion in experimental (upper panels) trials, or an equivalent duration in control (lower panels) trials. White (on) and black (off) bars below the abscissa indicate target visibility, which is highlighted by vertical dashed lines that cross the time series data. Overall group mean is represented by the thick red line in each panel.
Figure 6
 
Individual participant (P5) mean (n = 5 trials) smooth eye (left panel) and arm (right panel) velocity in oculo-manual conditions. Blocked-order and random-order experimental trials are represented as solid black and grey lines, respectively. Participant was pursuing targets accelerating at −12, 0, or 12 deg/s2 through a 1000 ms occlusion. White (on) and black (off) bars below the abscissa indicate target visibility, which is highlighted by vertical dashed lines that cross the time series data. Red solid lines represent the mean smooth eye (left panel) and arm (right panel) velocity in blocked-order control trials.
Figure 6
 
Individual participant (P5) mean (n = 5 trials) smooth eye (left panel) and arm (right panel) velocity in oculo-manual conditions. Blocked-order and random-order experimental trials are represented as solid black and grey lines, respectively. Participant was pursuing targets accelerating at −12, 0, or 12 deg/s2 through a 1000 ms occlusion. White (on) and black (off) bars below the abscissa indicate target visibility, which is highlighted by vertical dashed lines that cross the time series data. Red solid lines represent the mean smooth eye (left panel) and arm (right panel) velocity in blocked-order control trials.
Figure 7
 
Group mean reappearance eye and arm velocity in experimental trials as a function of condition, trial order, target acceleration, and occlusion (600 ms – upper panel; 1000 ms - lower panel). Ocular pursuit (Oc) is depicted by solid lines with filled squares, whereas ocular-manual pursuit (OcM) is depicted by solid lines with unfilled squares. Blocked-order and random-order trials are represented by black and grey lines, respectively. Arm velocity is depicted by broken lines with filled circles and follows the color coding for trial order. Target velocity (TGT) is shown for comparison and is represented by filled red circles.
Figure 7
 
Group mean reappearance eye and arm velocity in experimental trials as a function of condition, trial order, target acceleration, and occlusion (600 ms – upper panel; 1000 ms - lower panel). Ocular pursuit (Oc) is depicted by solid lines with filled squares, whereas ocular-manual pursuit (OcM) is depicted by solid lines with unfilled squares. Blocked-order and random-order trials are represented by black and grey lines, respectively. Arm velocity is depicted by broken lines with filled circles and follows the color coding for trial order. Target velocity (TGT) is shown for comparison and is represented by filled red circles.
Figure 8
 
Group mean total eye displacement (TED) in experimental trials as a function of trial order, target acceleration, and occlusion. Contributions from smooth eye displacement (SED) and saccadic eye displacement (SAD) are represented by filled grey and white bars, respectively. Error bars show the group standard deviation of TED. Target displacement (TGT) is represented by filled red circles
Figure 8
 
Group mean total eye displacement (TED) in experimental trials as a function of trial order, target acceleration, and occlusion. Contributions from smooth eye displacement (SED) and saccadic eye displacement (SAD) are represented by filled grey and white bars, respectively. Error bars show the group standard deviation of TED. Target displacement (TGT) is represented by filled red circles
Figure 9
 
Model of ocular pursuit receiving retinal and extra-retinal inputs. Retinal input is arranged as a negative visual feedback pathway. Extra-retinal input is received from either a direct or indirect (predictive) loop. When there is no expectation of future target movement, sw2 is set to allow the direct loop to operate, whereas when expectation is high (i.e., motion is predictable), sw2 is set to allow positive feedback via the short term memory (MEM) which holds a reference level of velocity captured from previous samples and can be used to predict target motion. The input to both direct and indirect pathways comes from sampling and holding a copy of the visuomotor drive signal (vmd) in module S/H. It is assumed that the gain β is modified by expectation (max value unity) and by conflict (or mismatch) detection. Sudden withdrawal of input causes a temporary reduction of gain β, so that eye velocity initially falls, but can then recover (e.g., towards the end of a transient occlusion). The inclusion of a more robust short-term store, MEM, within the indirect loop includes enables velocity information to be held over longer periods and during fixation. The gain, α(t), at the output of MEM can also be modulated as a function of time to allow more complex motion trajectories to be produced in anticipation of expected target motion (e.g., negative or positive acceleration) visual feedback delay – τv = 0.1 s.
Figure 9
 
Model of ocular pursuit receiving retinal and extra-retinal inputs. Retinal input is arranged as a negative visual feedback pathway. Extra-retinal input is received from either a direct or indirect (predictive) loop. When there is no expectation of future target movement, sw2 is set to allow the direct loop to operate, whereas when expectation is high (i.e., motion is predictable), sw2 is set to allow positive feedback via the short term memory (MEM) which holds a reference level of velocity captured from previous samples and can be used to predict target motion. The input to both direct and indirect pathways comes from sampling and holding a copy of the visuomotor drive signal (vmd) in module S/H. It is assumed that the gain β is modified by expectation (max value unity) and by conflict (or mismatch) detection. Sudden withdrawal of input causes a temporary reduction of gain β, so that eye velocity initially falls, but can then recover (e.g., towards the end of a transient occlusion). The inclusion of a more robust short-term store, MEM, within the indirect loop includes enables velocity information to be held over longer periods and during fixation. The gain, α(t), at the output of MEM can also be modulated as a function of time to allow more complex motion trajectories to be produced in anticipation of expected target motion (e.g., negative or positive acceleration) visual feedback delay – τv = 0.1 s.
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×