How does our visual system make sense of complex retinal input? While machine vision excels at tasks like object classification, it lacks biological insight. Conversely, classical visual models offer biological fidelity but limited scope. This project bridges this gap by studying how regulatory mechanisms in natural vision shape computational processing. We focus on regularization—constraints that optimize the input-output mapping, like energy efficiency in biology or noise robustness in AI. Our hypothesis is that these regulatory mechanisms play fundamental computational roles beyond mere maintenance. Through computational modeling, theoretical analysis, and experimental validation, we investigate how cellular-level regulatory processes influence visual cortex development and function. A key focus is understanding how retinal spontaneous activity and other regulatory mechanisms contribute to natural vision, with implications for both basic science and retinal prosthetics.
We focus on regularization—constraints that optimize the input-output mapping, like energy efficiency in biology or noise robustness in AI. Our hypothesis is that these regulatory mechanisms play fundamental computational roles beyond mere maintenance. Through computational modeling, theoretical analysis, and experimental validation, we investigate how cellular-level regulatory processes influence visual cortex development and function.