Representations in Rodent Primary Visual Cortex During Visual Discrimination

Although the hierarchical pathways underlying visual decision making in primates have been studied extensively, the neural circuits underlying visual decision making in rodents are not as well understood. Meanwhile, there is increasing evidence to suggest that while sensory cortices have been largely studied for their feedforward processing role, massive feedback projections distribute representations of nonsensory variables such as movement and choice across cortex, even dominating activity within early sensory cortices. The role of these representations is not well understood within a sensory discrimination paradigm. We therefore set out to investigate the representations and contributions of rodent primary visual cortex (V1) in a visual decision making task. In this thesis, I describe a novel visual discrimination task for freely moving rats. Subjects are presented with a set of distributed flickering dots (a “cloud of dots”), and are asked to judge whether there are more dots in the upper or lower visual hemifield. Subjects report their choice with a nosepoke into the corresponding water delivery port. To facilitate experimental access in this task, I developed a software-based approach to control viewing angle at the center port by reinforcing head position using closed-loop online video tracking. While subjects are able to find and stably carry out the comparison rule when it is necessitated by the statistics of the stimulus distribution over trials, behavioral experiments revealed that at baseline, subjects reliably converge on an abbreviated strategy, such that only half of the full stimulus is necessary and sufficient to drive the behavior. I used tetrode recordings to interrogate neuronal responses in primary visual cortex (V1) of behaving animals. In addition to classically responsive visual neurons, I report overlapping subpopulations of V1 single neurons that are not only responsive, but also selective to choice side and outcome. I show that despite a lack of clear structure to the organization of sensory and nonsensory information in V1 single neurons, the trial-to-trial variability of stimulus-driven activity depends on the behavioral relevance of visual information streams. I further quantified the intrinsic behavioral bias towards the lower visual field. I found that animals could adapt their strategy to the statistics of their stimulus environment over many trials, but at baseline overwhelmingly preferred to base their decisions on the lower visual field, even when this strategy was suboptimal. Preliminary optogenetic inactivation experiments showed that stimulus period V1 activity was necessary for accurate discrimination behavior, while outcome period V1 activity had little effect on next trial choice. Finally, motivated by the role of corticostriatal projections in the analogous auditory "cloud of tones" task, I explored the organization of the corticostriatal projection from V1. Here I observed an asymmetry in the pattern of projections from V1 to the striatum, coinciding with V1 mapping of visual space and the asymmetry in stimulus usage by our subjects