Background
When you haptically explore an object it is necessary to integrate information about the location of somatosensory input on the skin (touch) with information about the posture of the stimulated limb (proprioception). To investigate the time course of the integration between touch and proprioception, we analysed eye movements directed to somatosensory targets. That eye movements can be very informative about tactile encoding has been shown, among others, by Groh and Sparks (1996), who found that saccade trajectories to somatosensory targets on the hands are curved when the hands are crossed over the body midline. At the onset of the saccade only the somatotopic (skin) location is available and thus the saccade will be initiated towards that side of the body. When the somatotopic location is translated into an external frame of reference by integrating proprioception, the saccade direction is adjusted and the resulting saccade is curved. Thus, the curvature of saccades can inform us on the timing of integration of touch and proprioception.

Method
We measured eye movements to somatosensory targets on the hands in a crossed or uncrossed arm posture. In order to control the time of the onset of the saccade, we used a delayed response procedure in half of the trials. Thus, four blocks of trials were measured: crossed, no delay; crossed, delay; uncrossed, no delay; uncrossed, delay. The participants were seated with their arms either crossed or uncrossed and their ring fingers placed at marked locations just below the outer sides of a computer screen. Solenoid tappers were taped to the dorsal side of the distal phalanx of both ring fingers. The participant's head position was fixed by a chinrest, and a head-mounted Eyelink II® eye tracker recorded their eye movements. The arms and hands could not be seen by the participant. A trial started by the participant fixating a cross in the middle of the screen, followed by a stimulation of one of his ring fingers. The task for the participant was to make a saccade to the felt location of the stimulus as quickly and accurately as possible, after the go signal.

Results & Discussion
We analysed error rates, saccade trajectories, and saccade latencies. More errors were made when the hands were crossed, indicating the difficulty of the task while assuming this posture. We did not find any differences in the any of the properties of the saccades between the delayed crossed and uncrossed conditions and the no-delay uncrossed condition. Mean saccade onset latency was longer when the hands were crossed as compared to when they were uncrossed in the non delayed conditions. Participants made some curved saccades in the no-delay crossed condition, but they were less frequent than what was found by Groh and Sparks (1996). Yet, the few curved saccades that were found were made when the saccade latency was relatively short. This pattern may indicate that in the crossed-hands condition participants wait for the somatosensory input to be integrated with proprioception, before executing the saccade. This strategy (possibly learned) could help them to avoid errors in the execution of saccades in the crossed hand condition, but not in the uncrossed condition. Thus, the difference in saccade latency between crossed and uncrossed arm configurations is an indication of the timing of integration of somatosensory input with proprioception.

Reference
Groh, J. M., & Sparks, D. L. (1996). Saccades to somatosensory targets. I. behavioral characteristics. Journal of Neurophysiology, 75(1), 412-427.