The orientation of the visual axis to an object of interest generally recruits both the eyes and the head. Such combined eye-head gaze shifts frequently occur during ongoing eye and head movements but the programming and coordination of both motor systems in such a condition remains unknown. For saccades programmed during ongoing smooth pursuit in a head restrained condition it has been shown that the error induced by the programming delay of the saccade is corrected by an integration of the retinal slip (RS) and the position error (PE) sampled 125 ms before saccade onset (De Brouwer et al., 2002). In this study, we investigated how this error is corrected during head free 2D movement.
Subjects were seated in front of a 1-m distant tangential screen. They were instructed to pursue a sinusoidal target (Frequency [0.5 to 0.9 Hz]) moving along a straight line in 2D (Orientation [0 to 360°] and amplitude [20 to 25°] randomly chosen). Between 1.2 and 1.8s later, the target made a step to a random position on the screen (range [-30 to 30°] horizontally and vertically) and subjects were instructed to reorient gaze as fast as possible and to make active use of head movements in the task. Their head was totally free to move. The position of both eyes was recorded by a video-based recording device (200 Hz) and head position was recorded by active infrared marker tracking cameras (200 Hz). The gaze orientation was then reconstructed using a previously described calibration method (Ronsse et al., 2007) leading to 0.5° accuracy.
We analyzed the gaze shift that occurred after the step of the target. After a certain latency, a combined eye-head gaze saccade was initiated, followed by a slow VOR phase to account for the final head movement. Investigating how the gaze saccade was programmed, we found a strong correlation between saccade amplitude and the PE at saccade onset. We demonstrate that, the retinal PE of the target, measured 125ms before saccade onset was corrected by the predicted gaze displacement in the latency period, which resulted in better correlations (F(152,152)=3.994, p<0.001) of saccade amplitude with the pre-saccadic PE than with PE measured 125ms before. The gaze control system thus predicted the gaze displacement during the saccade latency from the available RS measured 125ms before saccade onset. This behavior was similar to what has previously been reported in head-restrained experiments.
In conclusion, we proposed that the gaze control system uses similar mechanisms to program head restrained and head unrestrained saccades. Both RS and retinal gaze PE are used to overcome the internal delays of saccade programming and to predict the future position of the target.

Supported by: FSR, FRSM, IAP VI/4 DYSCO, PRODEX (Belgium), ESA (EU), GB is supported by a Marie Curie fellowship (EU)