A parsimonious computational model of visual target position encoding in the superior colliculus

Abstract

The superior colliculus (SC) is a brainstem structure at the crossroad of multiple functional pathways. Several neurophysiological studies suggest that the population of active neurons in the SC encodes the location of a visual target to foveate, pursue or attend to. Although extensive research has been carried out on computational modeling, most of the reported models are often based on complex mechanisms and explain a limited number of experimental results. This suggests that a key aspect may have been overlooked in the design of previous computational models. After a careful study of the literature, we hypothesized that the representation of the whole retinal stimulus (not only its center) might play an important role in the dynamics of SC activity. To test this hypothesis, we designed a model of the SC which is built upon three well-accepted principles: the log-polar representation of the visual field onto the SC, the interplay between a center excitation and a surround inhibition and a simple neuronal dynamics, like the one proposed by the dynamic neural field theory. Results show that the retinotopic organization of the collicular activity conveys an implicit computation that deeply impacts the target selection process.

A unified dynamic neural field model of goal directed eye movements

“…Finally, a preference for orienting gaze shifts that bring the target near the center of the visual field was here assumed. Explanatory reasons for favoring such saccades can be computationally evaluated, for instance by adopting a retinotopic yet non-homogeneous representation of the visual field (with a greater accuity in the fovea, at the center of the retina), as developed in Taouali et al (2015). In addition to limiting the extent of the sensorimotor regularities to be learned, a saccade then also leads to the magnification of the target image.…”

“…Therefore, they are not in principle limited to the detection of stimuli defined by a single lateral connectivity profile (for an instance of complex preprocessing to abstract from raw input images, see Maggiani, Bourrasset, Quinton, Berry, & Sérot, 2016). Also, complex transformations and projections can be considered, for instance using alog-polar representation of the visual field (Taouali, Goffart, Alexandre, & Rougier, 2015) or relying on self-organizing maps (Lefort, Boniface, & Girau, 2011). While Taouali et al (2015) provide details on how the DNF approach can account for the encoding of target location in the deep superior colliculus, Gandhi and Katnani (2011) review the different plausible decoding mechanisms from a topologically organized population of neurons, focusing on saccade generation.…”

“…Also, complex transformations and projections can be considered, for instance using alog-polar representation of the visual field (Taouali, Goffart, Alexandre, & Rougier, 2015) or relying on self-organizing maps (Lefort, Boniface, & Girau, 2011). While Taouali et al (2015) provide details on how the DNF approach can account for the encoding of target location in the deep superior colliculus, Gandhi and Katnani (2011) review the different plausible decoding mechanisms from a topologically organized population of neurons, focusing on saccade generation. They compare the "vector summation" and "vector averaging" approaches, the latter being preferred in most of the DNF literature, as well as in the current paper.…”

“…Learning to foveate the target and keep it in the central visual field will reduce the problem to processing the motion signals and characterize the dynamics of the input stimulus. Also pointing in such direction, a model of target position encoding in the superior colliculus has been recently proposed in Taouali et al (2015), demonstrating that apparently complex foveation performance can be accounted for by dynamical properties of neural field and retinotopic stimulation (using a log-polar representation of the visual field).…”

A global framework for a systemic view of brain modeling

“…It is composed of several layers, some receiving mainly visual information from many regions in the brain, including directly from the retina. The more super cial sensory layers are topographic maps of the surrounding environment and are in direct association with deeper motor layers for eye movements towards the place elected by competition in the sensory layer (Taouali et al, 2015). It has been remarked that this structure can also perform direct sensorimotor associations for orientation of the whole body for tracking novel stimuli, for defensive movements and ight in case of a danger (Dean et al, 1989).…”