Eye–hand coordination during visuomotor adaptation: effects of hemispace and joint coordination

Abstract

We previously examined adaptive changes of eye–hand coordination during learning of a visuomotor rotation. Gazes during reaching movements were initially directed to a feedback cursor in early practice, but were gradually shifted toward the target with more practice, indicating an emerging gaze anchoring behavior. This adaptive pattern reflected a functional change of gaze control from exploring the cursor–hand relation to guiding the hand to the task goal. The present study further examined the effects of hemispace and joint coordination associated with target directions on this behavior. Young adults performed center-out reaching movements to four targets with their right hand on a horizontal digitizer, while looking at a rotated visual feedback cursor on a computer monitor. To examine the effect of hemispace related to visual stimuli, two out of the four targets were located in the ipsilateral workspace relative to the hand used, the other two in the contralateral workspace. To examine the effect of hemispace related to manual actions, two among the four targets were related to reaches made in the ipsilateral workspace, the other two to reaches made in the contralateral workspace. Furthermore, to examine the effect of the complexity of joint coordination, two among the four targets were reaches involving a direct path from the start to the target involving elbow movements (simple), whereas the other two targets were reaches involving both shoulder and elbow movements (complex). The results showed that the gaze anchoring behavior gradually emerged during practice for reaches made in all target directions. The speed of this change was affected mainly by the hemispace related to manual actions, whereas the other two effects were minimal. The gaze anchoring occurred faster for the ipsilateral reaches than for the contralateral reaches; gazes prior to the gaze anchoring were also directed less at the cursor vicinity but more at the mid-area between the starting point and the target. These results suggest that ipsilateral reaches result in a better predictability of the cursor–hand relation under the visuomotor rotation, thereby prompting an earlier functional change of gaze control through practice from a reactive to a predictive control.

Appendix

Assessments of adaptive changes, after-effects, and explicit knowledge

The participants underwent two types of pre-tests and three types of post-tests before and after the practice, respectively. Procedures of these tests were described in the main text (Table 1). All these tests were performed without online feedback, thereby excluding any changes attributed to the online visual feedback control. Relative changes of reaching direction between pre- and post-tests were used to assess the magnitudes of adaptive changes (adaptive shift), after-effects, and explicit knowledge (explicit shift) that reflected different types of knowledge of the visuomotor rotation acquired through practice (Heuer et al. 2013; Rentsch and Rand 2014; Sülzenbrück and Heuer 2011). The details of these assessments were similar to those described in our previous study (Rentsch and Rand 2014).

For data analysis, angular deviation of a vector (from the hand position at movement onset to that at movement offset) from a SP-target vector (from the starting position to the target) was computed in each trial of the pre-test 1, post-test 1 and post-test 2. In each trial of explicit pre-test and explicit post-test, angular deviation of the judged direction of the line from the direction of the SP-target vector was computed. Next, a circular mean (Berens 2009) of angular deviations across the first three trials of each test was calculated in each target direction and participant. Then, the difference between the circular mean of post-test 1 (or post-test 2) and that of pre-test 1 was calculated as adaptive shift (or after-effect), and the difference between the circular mean of explicit pre-test and that of explicit post-test was calculated as explicit shift.

Adaptive shifts reflect both implicit and explicit adjustments of reaching directions based on implicit and explicit knowledge of the visuomotor rotation acquired through practice (Heuer et al. 2013). Adaptive shifts of −75° would indicate that both types of adjustments combined fully compensate for the 75° rotation. As shown in Fig. 6a, mean adaptive shifts were substantially smaller than −75°, suggesting that online feedback-based adjustments were used aside from these two types of adjustments to compensate for the rotation. A 2 (group: IDE small vs IDE large) × 4 (target direction) ANOVA with repeated measures revealed no significant main effects (p > 0.05).

Fig. 6

Mean adaptive shifts (a), after-effects (b), and explicit shifts (c) are plotted against target directions (Fig. 1a, left plot). Mean values of all participants are plotted for the small-IDE group (filled columns) and the large-IDE group (open columns). The error bars represent standard errors

After-effects reflect only implicit adjustments of reaching directions based on implicit knowledge acquired through practice (Heuer et al. 2013). After-effects of -75° would indicate that participants acquired perfect implicit adjustments to compensate for the 75° rotation. Mean after-effects were largest for reaching to T1 and smallest to T4 (Fig. 6b). The 2 × 4 ANOVA showed a significant target direction effect [F(3,60) = 5.18, p < 0.01, η ²_p  = 0.206]. A planned contrast revealed that there were significantly greater after-effects for targets related to simple joint coordination (Fig. 6b, T1 and T3; Fig. 1a, left plot in the main text) than for those related to complex coordination [T2 and T4, F(1,20) = 12.04, p < 0.01, η ²_p  = 0.376]. After-effect tended to be greater for reaching to the ipsilateral targets (T1 and T2) than to the contralateral targets [T3 and T4, F(1,20) = 3.71, p = 0.068, η ²_p  = 0.157]. Note that the visual feedback was veridical for both pre-test 1 and post-test 2, thereby having no distinction between the target effects of hemispace-visual stimuli and hemispace-manual actions. There were no other significant main effects (p > 0.05).

Explicit shifts reflect the magnitude of explicit knowledge of the visuomotor rotation acquired through practice (Heuer et al. 2013). Explicit shifts of −75° would indicate that participants acquired perfect explicit knowledge of the applied 75° rotation. Even though mean explicit shifts were generally greater for the small-IDE group (Fig. 6c, filled columns) than those for the large-IDE group (open columns), there was large inter-individual variability (see error bars of Fig. 6c). Consequently, group effect and target direction effect were not significant (2 × 4 ANOVA: p > 0.05), whereas a group by target interaction just failed to reach a significance [F(3,60) = 2.36, p = 0.08, η ²_p  = 0.106].