Movement latency

2.4.1.1 Effect of task on mean latencies

Mean saccadic and manual latencies for eye and hand in the different tasks are given in Table 1.

Table 1 Mean eye and hand latencies and standard deviations between subjects for different tasks (each cell represents the mean of subject means, N=10)

Task eye hand

mean (ms) standard

deviation (ms) mean (ms) deviation (ms)standard

A steps (persisting target) 167 23 245 20

B steps (flashed target) 176 20 255 20

C proGap 160 39 254 48

D memory 318 64 382 99

E scanning 283 71 376 105

F antiGap 377 72 473 104

To determine the prevailing patterns for latencies, separate cluster analyses were performed on the single-trial data of eye and hand. Eye and hand latencies were found to be organised in the same two clusters: the steps conditions (persisting and flashed target) and condition proGap (A, B, C) formed one cluster, while conditions memory, scanning, and antiGap (D, E, F) formed the other cluster. The first cluster is characterised by lower latencies, the second by higher latencies. A plot of mean latencies shows these two clusters (cf. Figure 3).

hand eye task latency (ms) 0 50 100 150 200 250 300 350 400 450 500 550 steps steps (flashing) proGap memory scanning antiGap

Figure 3 Mean latencies of eye and hand in different tasks

To determine whether the differences between clusters and between the different movement types are significant, an analysis of variance was required. To detect any differential effects of the task on eye or hand, mean latencies were submitted to a 6x2 analysis of variance with the factors task (conditions A, B, C, D, E, F) and movement type (eye versus hand). There was a highly significant main effect for task (F= 26.11, df=5, p<.001), showing that mean latencies differed over conditions. This was also true for movement type (F=62.16, df=1, p<.001), showing that hand latencies were longer than eye latencies. No interaction of task with movement type was observed. Therefore, eye and hand were not differentially affected by the tasks applied. Planned comparisons confirmed the two latency clusters for eye and hand movements (F=146.19, df=1, p<.001).

In general, the primary saccadic eye movement (M=247 ms, sd=92 ms) started 84 ms before initiation of the hand movement (M=331 ms, sd=94 ms). This mean value is slightly larger than that obtained by other authors for purely reflexive tasks, e.g., 70 ms in a speeded aiming task requiring button pressing (Helsen et al. 1998), or 73 ms when quickly pointing to perturbed targets at 15° eccentricity (Carnahan and Marteniuk 1994).

2.4.1.2 Time interval between the end of the eye movement and the end of the hand movement

The eye arrived at the target 386 ms (sd=106 ms) before the hand. This time interval is long enough to permit correction of the limb position by visual information about the target location (e.g., Jeannerod 1988; Elliott and Allard 1985).

To check whether this time interval varies with the task, an analysis of variance with the task as within-subjects factor and the time interval between the end of the eye movement and the end of the hand movement as dependent variable was performed. It revealed a significant main effect for the task (F=2.90, df=5, p<.05). This time interval was found to be significantly larger under condition proGap than under the step condition with flashed target and under condition memory (both p <.05), as shown by a post hoc analysis, Tukey’s HSD. A closer inspection of the data revealed that this larger time interval was due to longer duration of the hand movement, which delayed the end of hand movement.

2.4.1.3 Correlation of ocular and manual latencies

To investigate temporal coupling of eye and hand, the correlations of eye and hand latencies were calculated on a trial by trial basis. Subsequently, the mean correlation for each task was calculated by averaging the z-transformed correlation coefficients of each subject. They were found to vary between the tasks from .32 to .74 (see Table 2).

Table 2 Mean trial to trial correlation and standard deviation between subjects of eye and hand latencies in different tasks (each cell represents the mean of the individual correlation coefficients, N=10)

Task pearson correlation deviation standard

A steps (persisting target) .49 .13 B steps (flashed target) .42 .30

C proGap .32 .19

D memory .68 .39

E scanning .59 .42

F antiGap .74 .34

(All correlations significantly different from 0 at the 1% level (two-tailed), according to a t-test performed on the z- transformed correlation coefficients of each subject)

The task clearly affected latency correlations of eye and hand. An analysis of variance of the z-transformed correlations of eye and hand latencies with task as within-subjects factor yielded a highly significant main effect for the task (F=5.18, df=5, p<.001). This effect can be specified according to the clusters obtained in the analysis of mean latencies. Planned comparisons of the step conditions (flashed and persisting target) and proGap with the conditions memory, scanning and antiGap confirmed the clusters obtained for mean eye and hand latencies (F=20.83, df=1, p<.01). Thus, eye and hand show closer temporal coupling for movements to remembered targets, scanning and anti- movements than for reactive movements.

In general, two different types of saccades can also be found within the antisaccade task, i.e., correct voluntary antisaccades and wrongly executed reflexive prosaccades. In the present study, this difference in movement types was also observed for hand movements. This provided an opportunity for analysing whether the differences found for different tasks also apply to different types of movements within the same task. Therefore, we performed a separate analysis of latency correlations for different types of movements within the antisaccade task. Eye-hand latency correlations for “wrong” prosaccades and hand movements oriented towards a physically present visual cue were compared with those obtained for correct antisaccades and hand movements towards the cognitively derived target. Trials in which hand and eye moved in different directions were excluded from this analysis.

This criterion was applied to 224 trials in which only the eye erroneously made a reflexive prosaccade, whereas trials with a wrong “pro”-movement only for the hand were absent.

Latency correlation for trials with wrong pro-movements was r=.50 (N=35, p<.01, two- tailed) as opposed to r=.74 (N=495, p<.001, two-tailed) for correct anti-movements. A comparison of the transformed correlation coefficients (Bortz 1993, p. 203; StatSoft 1999, paragraph “other significance tests”) showed that correlations for wrong pro-movements were significantly lower (p<.05) than for correct anti-movements.