gms | German Medical Science

Objective: Exploration and restoration of the function of neural circuits throughout the brain, including the retina, will be greatly aided by the ability to artificially evoke arbitrary patterns of neural activity in diverse cell types. Epiretinal electrical stimulation is a method with strong technical advantages for achieving this goal for retinal implants, but existing approaches have yet to achieve large-scale precise encoding of spatiotemporal firing patterns in multiple cell types. Here we demonstrate for the first time the ability to accurately reproduce large-scale, naturalistic firing patterns in two major RGC types in the macaque retina simultaneously using a bi-directional closed-loop approach.

Materials and Methods: Multi-electrode recording and visual and electrical stimulation were performed on nearly complete populations of ON and OFF parasol ganglion cells in isolated macaque retina using a 512-electrode system. Guided by an empirical calibration of electrical stimuli obtained during the experiment, electrical stimulation sequences tailored to precisely reproduce recorded neural responses to naturalistic visual stimuli were delivered to the retina.

Results: Across many stimuli, the electrically-evoked activity was consistently more similar to the target visual response than repeated trials of light-evoked responses were to one another. Further, image decoding performance using the electrically-evoked responses was comparable with that obtained using visually-evoked responses.

Discussion: The results together indicate that electrical stimulation can encode precise and relevant visual information in complete populations of neurons of distinct types, suggesting strong translational potential for neural interfaces of the future.

Acknowledgment: Supported by NSF Graduate Research Fellowship Grant No. 2146755, NSF Grant No. 1828993, Stanford Wu Tsai Neurosciences Institute, and NIH NEI R01-EY021271.