Spatial updating and the maintenance of visual constancy
The model parameters and internal states (remembered target position) are estimated sequentially using the EKF method.
The proposed model replicates two fundamental experimental observations: continuous gaze-centered updating of visual memory-related activity during smooth pursuit, and predictive remapping of visual memory activity before and during saccades.
Here, we propose a unified state-space model (SSM) to account for the dynamics of spatial updating during two types of eye movement; saccades and smooth pursuit.
Our proposed model is a non-linear SSM and implemented through a recurrent radial-basis-function neural network in a dual Extended Kalman filter (EKF) structure.
Such experiments have shown that humans and monkeys are able to do this despite intervening saccades (Hallett and Lightstone, 1976; Mays and Sparks, 1980), head-unrestrained gaze shifts (Munoz et al., 1991), smooth pursuit eye movements (Lisberger et al., 1987; Schlag et al., 1990; Herter and Guitton, 1998; Baker et al., 2003; Medendorp et al., 2003; Blohm et al., 2005; Klier and Angelaki, 2008), translational motion of the head/body (Angelaki and Yakusheva, 2009; Clemens et al., 2012), and torsional rotation of the head/body (Medendorp et al., 2003; Klier and Angelaki, 2008).
The detailed neurophysiological mechanisms are not yet known, but it is thought that this involves the use of internal efference copies of eye motion (Mays and Sparks, 1980; Sommer and Wurtz, 2006), and results in the recalculation of an appropriate motor saccade vector for the final eye position in areas such as the superior colliculus (SC) (Groh and Sparks, 1996).
Since experimental work is extremely difficult and often (in animal experiments) can only target selected areas and signals at one time, it is important to have a theoretical framework to guide such experiments.
Spatial updating is the ability to locate targets that were perceived before an intervening self-motion displaces the original sensory frame of reference (Baker et al., 2003; Klier and Angelaki, 2008; Sommer and Wurtz, 2008; Crawford et al., 2011).
In the oculomotor system, this is often studied using the “double step” task, where subjects are required to make an accurate saccade toward a remembered visual target after an intervening eye motion.
Thus, remapping, spatial memory, and sensorimotor updating may involve several common or inter-related features (Mays and Sparks, 1980; Duhamel et al., 1992; Walker et al., 1995; Batista et al., 1999; Tian et al., 2000; Nakamura and Colby, 2002; Balan and Ferrera, 2003a,b).
There is also reason to expect somewhat different mechanisms for updating visual space during behaviors that involve slow, continuous motion of the eye in space, like smooth pursuit eye movements or translations of the head (Angelaki and Yakusheva, 2009; Clemens et al., 2012; Dash et al., 2015).