Michael Beyeler

This paper develops a deep learning framework to control neural activity in the human visual cortex using a neuroprosthesis, achieving better performance than conventional methods.

This study benchmarks attention-based interpretability methods for Vision Transformers in retinal fundus imaging, finding gradient-weighted attention rollout provides the most faithful and biologically relevant attributions.

This paper identifies distinct head-eye coordination motifs during gaze shifts in freely moving mice and highlights HwE as an active coordination mode.

This study investigates how central and peripheral vision loss affect scene understanding and eye movements using gaze-contingent simulations, finding that peripheral loss more severely impairs scene semantics while central loss disrupts oculomotor behavior.

This paper introduces a computational virtual patient pipeline to predict human perceptual behavior with retinal prostheses, finding that preserving natural-image representations better reproduces human perception than adapting to phosphene-specific features.

This paper introduces a multimodal recurrent neural network to predict V1 activity in freely moving mice by integrating visual input and behavioral dynamics, showing state-of-the-art performance and revealing mixed selectivity in cortical function.

The paper presents a biophysically detailed model of retinal degeneration that simulates network-level responses to light and electrical stimulation, showing how degeneration affects RGC activity.

The paper reviews nonnegative sparse coding in the brain, its role in neural computation across regions, and its potential underlying mechanisms.

This study predicts retinal implant users' phosphenes using a model simulating individual retinal ganglion axon pathways, accurately replicating percept shapes.

The paper develops a computational model to study axonal stimulation in epiretinal visual prostheses and finds that activation thresholds vary with stimulus duration and electrode-retina distance.

This study uses a computational model to recommend optimal electrode placement for retinal implants to reduce axonal activation and improve visual perception.

pulse2percept simulates the perceptual experience of retinal prosthesis patients to aid in the development of visual prostheses.

This paper reviews the role of cortical plasticity in visual prosthetics, discussing how biological constraints affect the ability to restore functional vision through retinal implants.