Event-based softcore processor in a biohybrid setup applied to structural plasticity

Abstract

The goal in neuromorphic engineering is to design circuits and systems which emulate the computational principles of biological nervous systems. As these circuits follow the same fundamental design principles as their biological counterparts, they represent an elegant solution for the design of bio-hybrid computing architectures. We present a neuromorphic bio-hybrid system in which electronic circuits are coupled directly to neuronal cell cultures for providing low-level access to biological signal processing. To form this bio-hybrid, we introduce a backbone system which allows the implements of different network topologies by routing Address-Event Representations (AER) of spikes from biological cells to neuromorphic circuits and vice versa. The use of a soft-embedded-processor realtime system allows the exploration of topology-evolving setups. The final goal in the application of this backbone system is the creation of a bio-hybrid neural network that is structurally plastic. Preliminary results have already been obtained in a first verification step, successfully forming a bio-hybrid neural network and thus promising a novel approach towards bioinspired Brain-Machine Interfaces.