Polychrony and the process view of computation

Abstract

Some realistic models of neural spiking take into account spike timing, yet the practical relevance of spike timing is often unclear. In Eugene Izhikevich's model, timing plays a crucial role by allowing for the natural formation of polychronous circuits. In such circuits, individual elements may figure in a number of distinct assemblies, their role in each determined by their timing relative to other neurons. I show that this reflects a distinct organizational principle from notions of pluripotency, redundancy, or reuse and argue that properly understanding this phenomenon requires a shift to a time-sensitive, process-based view of computation.