Neural computation without neurons

Our model for learning is based on neural networks, i.e., networks of linear threshold devices. Even physical realizations of neural computation, such as molecular or electrical circuits, effectively mimic these network architectures at an element-by-element level. Here, we explore an alternative paradigm for neural computation - inevitable collective physical processes that can learn in Hebbian-inspired ways without being designed to mimic a neural network element-by-element. Through theory and experiment, we show how nucleation in molecular systems can learn to recognize complex patterns in chemical or mechanical stimuli. Our work suggests that exploiting ubiquitous physical phenomena, such as nucleation, is an underexplored powerful route to neural computation without neurons.