Model-based separation of visual and oculomotor signals in marmoset lateral intraparietal area

Abstract

The lateral intraparietal area (LIP) integrates visual information with task demands to guide saccadic eye movements. Most prior studies have relied on imposed prolonged delays to separate sensory and motor activity, whereas natural behavior involves rapid sensory-motor transformations. To examine LIP function during more naturalistic oculomotor behavior, we recorded multi-unit activity from LIP in two marmosets performing a task requiring rapid selection of salient visual stimuli. Because sensory and motor signals temporally overlap in such fast tasks, we used a Generalized Linear Model (GLM) that separately captured visually driven and oculomotor-related components. The model explained substantial neural variance during correct and missed saccades, as well as spontaneous eye movements, and revealed heterogeneous sensory and motor encoding across recorded neural units. Our approach demonstrated that the origins of neural activity can be segregated without relying on imposed delays, and showed the presence of a strong post-saccadic motor signal in LIP.