Fig. 1. Introduction to PNCA.

a Cellular Automata (CA) consist of computational units called cells, which update states according to interactions with neighboring cells. These microscopic local cell interactions can lead to emergent phenomena such as self-organization at the macroscopic scale, and even a global state agreement. b Neural Cellular Automata (NCA) encode the local update rules for CA using artificial neural networks and can be trained using modern deep learning techniques to perform tasks, such as image classification through collective agreement of cells. c Photonic Neural Cellular Automata (PNCA) directly implement NCA in physical systems by harnessing the speed and interconnectivity of analog photonic hardware, which includes linear operations via light interference and nonlinear activations via nonlinear optics. This endows photonic neural networks with the robust, reliable, and efficient information processing capabilities of NCA, hence overcoming several practical challenges facing light-based computing