Novel ion-gel nanofiber-based analog neuromorphic substrates for emulating neural networks with ephaptic interactions

Design and build a new type of neuromorphic substrate for analog computation: a deformable, hydrogel-based or ion-gel nanofiber-based networks with toroidal connectivity, enabling persistent, self-sustained spatial computation reminiscent of neural dynamics (e.g., traveling waves, continuous attractors). (Toroidal geometry is critical to avoid boundary effects). The system would exploit the material properties of ion-gel nanofiber networks and hydrogels—continuous deformation, ionic diffusion, and nonlinear response—to implement computation directly in the physical substrate. Activity could be read out via embedded sensing (e.g., electrical or optical signals), and coupled to a learning mechanism inspired by physical error backpropagation (à la PL McMahon) to train it to perform nontrivial tasks such as perception, pattern recognition, or closed-loop control in real time. (This research project will need a great deal of artistic creativity!)