Multiple polymer memristors with homeostatic plasticity through ionic global gating

Abstract

Organic neuromorphic hardware has attracted significant interest because it readily offers excellent biocompatibility and supports biomimetic ionic signaling. However, emulating a combination of Hebbian plasticity with positive feedback and homeostatic plasticity for brain-like ensemble stabilization remains challenging at the multi-device level. To address this, we developed an ionic gating strategy employing a lateral global gate. The global gating triggers ion redistribution, enabling synchronous adjustment of the memristive strength across multiple polymer memristors. A small network of these memristors exhibits homeostatic global regulation while preserving functional differences among individual synaptic weights. Moreover, the simulations of an L1-constrained neural network with mapping of the device data demonstrate that an appropriate global weight budget is essential for optimal learning performance, indicating the superiority of incorporating a homeostatic stabilization mechanism in neuromorphic circuits.