Active vision across scales: internal signals shape visual processing from retinal circuits to cortical population dynamics

Rather than conceptualizing vision as a passive feedforward transformation of external stimuli, the active vision perspective emphasizes that sensory input is continuously shaped by motor, autonomic, and internal brain processes. Neural activity in the visual system therefore reflects not only the structure of the external world, but also internally-generated signals that regulate how sensory information is sampled, filtered, and represented. These signals can act peripherally, by directly affecting the sampling of the environment, or centrally, by modulating, for example, the cortical population dynamics underlying sensory coding. In this thesis I examine both levels of interaction: first, I describe how an autonomic reflex such as the pupillary light reflex can shape visual responses in the early visual system by directly driving retinal responses. I then move to the visual cortex, where I studied how fluctuations in global internal states, such as arousal, dynamically reorganize the geometry of different cortical population during sustained sensory drive. By bridging peripheral and cortical levels, this thesis highlights how visual processing is continuously shaped by centrally generated signals that operate across hierarchical stages of the visual system.