Visuomotor velocity transformations for visually-guided manual tracking

Guillaume Leclercq, Gunnar Blohm, Philippe Lefevre


To achieve accurate visually guided arm movements the brain transforms visual input into appropriate motor commands for the arm. For reaches towards static targets this transformation accounts for the complete 3D eye-head-shoulder geometry (Blohm and Crawford 2007). However, position and velocity signals are processed by different neural pathways. Therefore, we ask whether a similar visuomotor transformation is also performed for velocity signals.

To address this question, we designed a dual quaternion model describing the complete visuomotor transformation geometry for pointing, accounting for 3D eye-in-head and head-on-shoulder rotations and translations. The model predicted compensation for (1) head roll and resulting counter-roll eye movements and (2) for false ocular torsion generated by a misalignment between the retinal and spatial coordinates during oblique gaze positions.

We tested these predictions on human subjects that performed manual tracking movements towards moving targets in darkness under different eye and head positions. To test prediction 1, subjects first had to roll their head towards either shoulder.  Then, they pointed to the central target, which started moving 1s later towards either the left or right with a angular vertical component of -10, 0 or 10 degrees. Subjects had to track the moving target with their hand while maintaining fixation. Testing prediction 2 was similar, but now the head was maintained in an upright position and subjects instead fixated oblique targets while the same tracking task is carried out.  We measured eye, hand and head movements and computed arm velocity during the open-loop period (first 200ms after movement onset).  This initial movement direction was then compared to the model predictions to check when the 3D eye-head-shoulder geometry was fully, partially or not at all taken into account in the visuomotor transformation.

Results showed that subjects compensated for head roll (R = 0.93, t(1533) = 102, p < .001) and the ocular counter-roll (R = 0.95, F(2,1532) = 7137 , p < .001 ) and also accounted for false torsion (R = 0.26, F(2,622) = 46,  p < .001). This suggests that for manual tracking movements, the brain achieves a visuomotor velocity transformation accounting for the complete 3D eye-head-shoulder geometry.