UCSB associate professor of computer science Michael Beyeler and his colleagues used a deep-learning model to design patterns of electrical stimulation for electrodes temporarily implanted in the visual cortex — the region at the back of the brain that processes visual information — of a blind participant.