Michaela Vystrčilová

The paper develops a spatial contrast model to capture nonlinear spatial integration in primate retinal ganglion cells using natural and artificial stimuli, showing improved performance over linear models.

This study examines how primate retinal ganglion cells adapt their spatial processing to natural versus artificial stimuli, finding that some cell types adjust receptive fields and surrounds, but efficient coding only partially explains these changes.

This study uses convolutional neural networks to investigate wide-field inhibition in the marmoset retina, revealing hidden computational aspects of retinal processing through interpretable models.

openretina provides a standardized framework for training and evaluating retinal models across datasets and species, revealing unexplained variance in current models.

This study compares predictive models for retinal ganglion cell responses, finding that CNNs outperform LN models, while LN models show better cross-stimulus generalization.