Identifying High-resolution Spatiotemporal Components Contributing to the Fast Spiking Response Dynamics of Visual Neurons

Abstract

In many brain areas, responses to sensory stimuli vary due to other cognitive, motor, or task factors. In the visual system the integration of these modulatory factors and visual stimuli can change the spatiotemporal characteristics of visual neurons at various spatial and temporal scales. High resolution changes in the neurons’ spatiotemporal sensitivity happening on fast timescales, however, can challenge computational models that aim to capture the neural computations underlying these fast dynamics. The time-varying visual sensitivity around the time of eye movements is an exemplar of such fast, dynamic modulatory computations. This study develops a statistical framework for identifying the high-resolution spatiotemporal components of visual neurons in the middle temporal area of macaque monkeys during a rapid eye movement task. The identified components can be used in building dynamic encoding models capable of characterizing the time-varying stimulus-response relationships with high resolutions and at the level of single-trial spiking activity. Such dynamic models with high temporal precision can be used to provide higher accuracy in the decoding of time-varying visual information from neuronal responses, which can in turn advance visual brain-machine interface systems to be able to operate robustly and with high accuracy in dynamic scenes.