Post

New in Neuron 👁️🧠🧪: Can AI help a cortical visual prosthesis control what the brain actually does after electrical stimulation? We used a bidirectional implant to model and shape stimulation-evoked activity in human visual cortex. doi.org/10.1016/j.ne... #UCSB #AI #BionicVision #Neurotechnology

First problem: the same electrical stimulation does not always produce the same neural response. Responses were much more variable across days than within a day, motivating a model that accounts for the brain’s ongoing state rather than treating stimulation as a fixed input-output mapping.

So we trained a deep forward model to predict the population response to multielectrode stimulation. On held-out days, it outperformed 1-to-1, linear, nonlinear, and dictionary-based models across multiple measures of prediction accuracy.

Prediction is useful. Control is the harder problem. Given a desired neural response, we used either gradient optimization or an inverse neural network to find a stimulation pattern predicted to produce it, then tested those patterns in vivo. Both beat conventional approaches.

But not every neural response is equally achievable... Evoked activity occupied a low-dimensional neural manifold. Targets farther from that manifold were much harder to reproduce (r = 0.85). The optimized methods also achieved their targets using lower stimulation currents.

Finally, the neural response mattered for perception. Recorded population activity predicted whether a phosphene was seen, and its brightness and color, substantially better than stimulation parameters alone. The emerging picture: stimulation → neural response → perception #BCI #Neuroscience