Content area
Biological neural systems evolved in a spacetime universe, inherently functioning across integrated spatiotemporal domains. Consistent with this reality, two fundamental biological processes---motor control and sensory perception---are intrinsically temporal. Yet visual processing, the dominant sensory system for diurnal species, is frequently mischaracterized: falsely linked to a camera and assumed to employ a purely spatial code, where spatial patterns are represented solely through topographic activation of neurons.
Yet this perspective overlooks a fundamental reality: neurons throughout the visual hierarchy exhibit pronounced sensitivity to temporal change. Concurrently, incessant eye movements (and head movements in some species) ensure retinal images are never static. This dissertation advances an alternative framework---active spatiotemporal coding---investigating how human oculomotor behaviors transform spatial patterns into spatiotemporal signals to facilitate visual processing.
Chapter 2 investigates how saccade-drift cycles modulate visual sensitivity during steady scene viewing. We demonstrate that these oculomotor patterns transform static scenes into dynamic retinal luminance flow, driving eccentricity-dependent coarse-to-fine processing: finer processing centrally, scaling coarser peripherally.
Chapter 3 reveals how smooth pursuit facilitates visual processing of moving objects. Surprisingly, idiosyncratic retinal motion patterns persist identically during fixation and pursuit. Retinal drifts and saccades during pursuit complementarily whiten natural scene power: drifts equalize signal strength to higher spatial frequencies, while saccades amplify signal amplitudes. Faster drifts during pursuit than fixation boost low-frequency power but attenuate high-frequency power, correspondingly enhancing low-frequency sensitivity while impairing high-frequency sensitivity. This indicates that retinal motion during pursuit is actively maintained for computational benefit---not minimized as “error”.
Chapter 4 uncovers how blinks reformat retinal input for active visual computation. We establish that blinks implement active spatiotemporal coding, facilitating visual processing. Notably, visual sensitivity improves when a blink---whether voluntary or reflexive---occurs during stimulus presentation. This enhancement is replicated by simulated blinks and arises from blink-induced luminance transients. Importantly, the effect is selective for low spatial frequencies. These findings address a longstanding evolutionary puzzle: why humans blink more frequently than is necessary for ocular lubrication, despite the potential visual disruption.
