Millions of people worldwide face partial or total vision loss due to inherited photoreceptor degenerative diseases, which currently have no cure. Retinal prostheses have been developed to restore vision by electrically stimulating surviving retinal neurons, but have low spatial resolution and nonselectively stimulate retinal ganglion cell (RGC) axons along with somata. We propose a biomimetic solution: using the neurotransmitter glutamate to chemically stimulate RGCs to avoid the disadvantages of electrical stimulation. Our results demonstrate that glutamate stimulation has a spatial resolution comparable to current-generation electrical prostheses, can stimulate RGC somata without stimulating axons, and can produce spatially differential responses in RGC subtypes. These results highlight the benefits of a neurotransmitter-based retinal prosthesis over current-generation electrical prostheses.