Retinal implants are used to replace lost photoreceptors in blind patients suffering from retinopathies such as retinitis pigmentosa. Patients wearing implants regain some rudimentary visual function. However, it is severely limited compared to normal vision because non-physiological stimulation strategies fail to selectively activate different retinal pathways at sufficient spatial and temporal resolution. The development of improved stimulation strategies is rendered difficult by the large space of potential stimuli. Here we systematically explore a subspace of potential stimuli by electrically stimulating healthy and blind mouse retina in epiretinal configuration using smooth Gaussian white noise delivered by a high-density CMOS-based microelectrode array. We identify linear filters of retinal ganglion cells (RGCs) by fitting a linear-nonlinear-Poisson (LNP) model. Our stimulus evokes spatially and temporally confined spiking responses in RGC which are accurately predicted by the LNP model. Furthermore, we find diverse shapes of linear filters in the linear stage of the model, suggesting diverse preferred electrical stimuli of RGCs. The linear filter base identified by our approach could provide a starting point of a model-guided search for improved stimuli for retinal prosthetics.

Integrating data from multiple experiments is common practice in systems neuroscience but it requires inter-experimental variability to be negligible compared to the biological signal of interest. This requirement is rarely fulfilled; systematic changes between experiments can drastically affect the outcome of complex analysis pipelines. Modern machine learning approaches designed to adapt models across multiple data domains offer flexible ways of removing inter-experimental variability where classical statistical methods often fail. While applications of these methods have been mostly limited to single-cell genomics, in this work, we develop a theoretical framework for domain adaptation in systems neuroscience. We implement this in an adversarial optimization scheme that removes inter-experimental variability while preserving the biological signal. We compare our method to previous approaches on a large-scale dataset of two-photon imaging recordings of retinal bipolar cell responses to visual stimuli. This dataset provides a unique benchmark as it contains biological signal from well-defined cell types that is obscured by large inter-experimental variability. In a supervised setting, we compare the generalization performance of cell type classifiers across experiments, which we validate with anatomical cell type distributions from electron microscopy data. In an unsupervised setting, we remove inter-experimental variability from data which can then be fed into arbitrary downstream analyses. In both settings, we find that our method achieves the best trade-off between removing inter-experimental variability and preserving biological signal. Thus, we offer a flexible approach to remove inter-experimental variability and integrate datasets across experiments in systems neuroscience. Code available at https://github.com/eulerlab/rave.