Neuronal mechanisms underlying training-induced vision recovery

Occipital strokes cause permanent damage to primary visual cortex (V1), the gateway for visual information processing in humans. Such patients lose conscious vision in the contralateral visual field, termed cortically- induced blindness (CB), although unconscious visual processing sometimes remains in the blind field, known as

blindsight. Importantly, visual discrimination training with moving stimuli placed just inside the blind-field border can restore some conscious vision in CB patients. However, overall visual capacities in the blind field remain highly impaired, making daily activities like reading and driving difficult or impossible. A number of neuronal

pathways linking the visual thalamus (dorsal lateral geniculate nucleus, LGN) with extrastriate cortex directly, or indirectly through residual portions of V1, have been proposed to underlie blindsight and training-induced vision recovery, respectively. However, no direct measures of changes in the neurons and circuits connecting LGN,

perilesional V1, and extrastriate cortex have ever been made after V1 lesions, especially simultaneously, and such invasive measures are not possible in humans. Here, we aim to fill this substantial gap by characterizing neuronal and circuit adaptations post-V1 lesions in a novel animal model of CB (Aim 1). We will then uncover

reconfigurations that occur in these neurons and circuits following training optimized to attain vision restoration (Aims 2-4). Our overarching goals are to provide a mechanistic explanation for both blindsight and training- induced vision recovery, and to provide a foundation for improving current vision restoration therapies throughout

the blind field. We will develop a CB model in ferrets, highly visual carnivores with early visual parallel processing streams and motion-selective extrastriate cortical areas homologous to those in primates. Ferrets have a large binocular visual field, unlike rodents, and can be trained to perform complex visual discrimination tasks, like

humans and primates. Yet larger cohorts of ferrets can be tested and trained across a battery of stimulus para- digms than is possible in primates. Ibotenic acid lesions of V1 will be made to create vision loss in the central ~50˚. Multi-electrode arrays will be inserted simultaneously into 3 critical nodes in the residual visual system:

LGN, perilesional V1, and the motion-selective postero-medial lateral suprasylvian (PMLS) area to test how direct LGN-PMLS and indirect LGN-perilesional V1-PMLS circuits are modified post-lesion (Aim 1). Separate cohorts of V1-lesioned ferrets will then be trained on custom regimes (Aim 2) inspired by neuronal response

properties measured in Aim 1. This will allow assessment of neurophysiological adaptations induced by optimal training paradigms (Aim 3). Finally, retrograde tracing and neurochemistry will be used to determine the contri- butions of distinct cell types and neuronal circuits to passive plasticity (Aim 1) versus training-induced vision

recovery (Aim 4). Together, these results will provide the most comprehensive picture yet of the neuronal mech- anisms engaged by V1 lesions, and how these differ between blindsight and training-induced visual recovery. This creates a critical scientific platform for future optimization of vision restoration in human CB patients.