Velocity tracking control best explains the modulation of saccade kinematics during eye-hand coordination

Fast movements like saccadic eye movements that occur in the absence of sensory feedback are thought to be controlled by internal feedback. Such internal feedback provides an instantaneous estimate of the output, which serves as a proxy for sensory feedback, that can be used by the controller to correct deviations from the desired plan. In the predominant view, the desired plan/input is encoded in the form of a static displacement signal (endpoint model), believed to be encoded in the spatial map of the superior colliculus (SC). However, recent evidence has shown that SC neurons have a dynamic signal that correlates with saccade velocity, suggesting that information for velocity-based control is available for generating saccades. Motivated by this observation, we used a novel optimal control framework to test whether saccadic execution could be achieved by tracking a dynamic velocity signal at the input. We validated this velocity tracking model in a task where the peak saccade velocity was modulated by the speed of a concurrent hand movement independent of the saccade endpoint. A comparison showed that in this task, the velocity tracking model performed significantly better than the endpoint model. These results suggest that the saccadic system may have additional flexibility to incorporate a velocity-based internal feedback control when imposed by task goals or context.