A: hand only B: eye only C: eye and hand D: rest
Still Moving Moving Still
Moving Still Moving Moving
Active Replayed Active Replayed
Fixation Tracking Tracking Fixation
Pursuit tracking At rest Compensatory tracking At rest
A: rest B: eyes only
At rest At rest
Table 1 Experimental conditions and visual stimuli. The white target was the target for all eye movements. The red target was the target for the green cursor, which was either controlled by the
subject using a computer mouse or was driven by the trajectory followed by the subject in a previous trial, which was stored and replayed
not be monitored during scanning, but was recorded in subject RCM and in other subjects outside the scanner during the same tracking tasks using the ASL 501 infrared reflectometry eye- tracking system, monitoring left eye position in two dimensions.
Subjects alternated every 17.6 s between four tasks in an ABCD CBAD sequence. A 22 factorial design was used, with main factors of hand tracking and eye tracking. Thus, task A in- volved pursuit manual tracking (“hand-only”), using the mouse to superimpose the green cursor on the moving red target while the eyes remained fixed on the stationary, central, white target (Fig. 1B). Task B involved ocular tracking of the moving white target (“eyes-only”); subjects held the mouse still, but followed the smoothly moving ocular target with their eyes. Task C was to pursue the moving white target with the eyes, while also tracking the moving green cursor using the mouse (“eye and hand”). The baseline task D was ocular and hand fixation (“fixation”); the white ocular target was displayed stationary at the centre of the screen, and the subjects held the mouse still.
While a 22 factorial design is relatively common, we used pursuit manual tracking in the hand-only (task A) condition, but used “compensatory manual tracking” (Poulton 1974) in the eye and hand condition (task C). In the latter case, the cursor was dis- placed from the central, stationary red target by the same pseudo- random waveform used to move the ocular target, and the subject was required to make compensatory movements with the comput- er mouse to return the cursor to the stationary target.
and hand spatially separated on the retina. Third, we recorded tracking errors and the path-length of the hand movement in the two conditions, and these were not significantly different.
Finally, to maintain equivalent retinal input in the four condi- tions, we used a “record and playback” technique to move the cur- sor on the screen during tasks without active hand movement (Table 1). Thus, the movement of the cursor was stored during task A and replayed in task B. The cursor movement was stored in task C and replayed in task D, etc. In this way, the movement of red target and the green cursor across the retina were approxima- tely similar in all tasks, while the white target was foveated in all tasks.
Note that, during successful performance of the combined eye- hand tracking task, the spatial and temporal trajectories of the eye and hand were the same (see Figs. 1B and 2C). In other words, knowledge that the eye was moving at a particular speed in a par- ticular direction, say to the left, could be used to guide the motion of the hand to the left: there was therefore the possibility of using eye and hand inter-communication to improve performance. It is this spatially and temporally consistent movement of hand and eye that we are considering to be the principal element of co-ordination within these experiments. Even in the other tasks, there may still be co-ordination, for example within the multiple joints of the arm, but this should be equivalent across the different tasks and, thus, not be expected to contribute to the functional activation observed.
This difference in tracking modalities can be justified as fol- lows. The factorial design requires a condition in which the eyes move without hand movement, which is straightforward, a condi- tion in which there is combined eye and eye tracking, which could have been achieved with normal pursuit of a moving target, and a condition in which there is hand movement without eye move- ment. A standard method to achieve this last condition is to use a “compensatory tracking” mode. Here, the eye can be shown to fix- ate the stationary target, while compensatory hand movements are used to control the cursor (Weir et al. 1989). However, normal pursuit tracking is markedly easier than compensatory tracking (Poulton 1974; Weir et al. 1989), and thus one could not simply compare the two. The difference in difficulty is largely because a compensatory display does not allow the subject to assess the po- sition and velocity of the target waveform, but only the instanta- neous tracking error and error velocity; in pursuit, one has access to target position and velocity cues (Weir et al. 1989)
Thus, we needed two comparable tasks involving hand track- ing, one with and one without eye tracking, in which cues about target velocity and position were equally available, in which equivalent retinal inputs were provided in each condition by equal spatial separation of the eye and hand targets on the screen in both conditions, and in which the manual tracking was of approximate- ly equal difficulty. Our design achieved all three of these aims (Fig. 1B): we used pursuit tracking with the hand while the eye was fixating a stationary target (task A, Table 1), so that the posi- tion and velocity of the target motion was available to the subject, albeit on the peripheral retina. Second, we used compensatory tracking with the hand whilst the eye was pursuing the target (task C), so that again target position and velocity cues were available from the ocular-motor system, while keeping the targets of the eye
Experiment two. Cerebellar activation during visually guided eye movements
Ocular tracking responses detected in experiment 1 proved to be only just above statistical threshold (P=0.01). Hence, in a second experiment, we contrasted activation within the cerebellum during only two conditions, ocular tracking of the moving target and ocu- lar fixation. As in experiment 1, ten axial slices were taken span- ning the vertical extent of the cerebellum, excluding the dorsal ce- rebral areas. Repetition rate was 4.4 s, and 128 images were ac- quired per session. This allowed us to double the number of statis- tical comparisons from 32 images per condition (experiment 1) to 64 per condition.
In this experiment, subjects followed the same target wave- form used in experiment 1, tracking the moving white ocular tar- get for 17.6 s and alternating with fixation of the stationary central target. Each task lasted 17.6 s and ran in an ABAB sequence. The same imaging parameters, slice locations etc. were used as before.
Experiment one. Cerebellar activation during co-ordinated eye and hand tracking
Tracking was performed under three conditions, either manual tracking alone, ocular tracking alone, or com- bined manual and ocular tracking.