A displacement and velocity based dual model of saccadic eye movements best explains kinematic variability

Abstract

Noise is a ubiquitous component of motor systems which leads to behavioral variability of all types of movements, including saccadic eye movements. Nonetheless, systems-based models of saccadic eye movements are deterministic and do not explain the observed saccade variability, only their central tendencies. Using stochastic models, we studied the variability in saccade behavior to test and distinguish between previously proposed deterministic saccade models. For this, the inter-trial variability in saccade displacement trajectories of human subjects was quantified while they performed repeated saccadic eye movements to a peripheral target. Based on fits to the data, we showed that existing models based on either displacement or velocity failed to capture the observed patterns in the variability of saccade trajectories. However, the observed behavior was captured by a dual control system, using a combination of displacement and velocity signal. The proposed model fits the mean displacement trajectory as well as the existing deterministic models. Taken together, our results suggest that the saccade system uses both desired displacement and velocity information. New and NoteworthyWe studied saccade behavior with a focus on the variability of the saccade trajectory. A stochastic model of the saccade system suggests that a dual control involving the control of displacement and velocity explains saccade behavior better than previously proposed models that utilize only displacement or velocity information. Our study resolves previous ambiguity regarding the use of displacement or velocity signals to guide saccades and provides a natural explanation for neural recordings that indicate multiplexing of displacement and velocity related information in the firing activity of neurons in the superior colliculus, a critical node in the oculomotor network that codes for saccadic eye movements.