Fig. 2 Typical hand and eye recordings from subject RCM during the four different conditions (A–D) used in ex- periment 1. The upper panels show the horizontal component of the target trajectory (fine line) and the cursor trajectory (thick line); the lower panels show the horizontal eye move- ments. The large upward de- flections in these traces repres- ent blinks. Units of horizontal motion are in screen pixels
Recordings of eye movements were made outside the scanner, using the same target waveform and with the same angular deviation and velocities of target; the tar- gets were displayed and viewed directly on a VGA com- puter monitor, although intensity and contrast were set approximately equal. Tracking was performed with a mixture of smooth pursuit and frequent small saccades. There were no gross differences between the eye- movement traces in the eye-only and eye-hand condi- tions (Fig. 2A, C).
The subjects’ manual tracking errors and mouse path- length were recorded during the scanning sessions and demonstrated that approximately equal tracking perfor- mance was achieved in the hand-only and the eye-hand conditions. We recorded the total movement of the cur- sor every 4.4 s (Fig. 3A) and also the positional distance of the cursor from the target. The distance moved by the mouse was 4% greater in the hand only condition than in the eye-hand condition (F1,10=2.84, P=0.12, ANOVA). Due to a programming mistake, we saved the positional distance between the cursor and the ocular target in the eye-hand tracking conditions and the error between the cursor and the manual target in the hand-only condi- tion. However, we were able to reconstruct the cursor-to- ocular target “fixation error” for both conditions by mea-
suring the errors in the record and playback conditions, and this allowed another comparison of the manual performance in the two manual tracking conditions (Fig. 3B). Again, there were no significant differences between the two tasks (fixation errors were 8% lower in the hand only condition than in the eye-hand condition, but this was not statistically significant: F1,10=3.40, P=0.09, ANOVA).
There is a strong periodic pattern to the path-length and tracking-error data in Fig. 3 because the pseudoran- dom target waveform varied in its speed and direction. The target trajectory was reset to the start of the wave- form every 17.6 s; the waveform repeat rate was 15 s, so that the waveform and scan acquisition would, unless re- set, synchronise only once every 100 s. By resetting the waveform, all functional scans were performed under similar task-difficulty conditions. However, it was then possible that the subjects might have improved their per- formance over time, as they could have acquired knowl- edge of this repeated target function. This did not appear to be the case; Fig. 3 indicates similar scatter of the er- rors and mouse path-length across the 128 scans (dura- tion 9.4 min); the slopes of regression lines through these data sets were not significantly different from zero (P>0.6).
Visual tracking using the hand-held computer mouse ac- tivated with very high significance cerebellar regions