| Abstract: 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. In this framework, inevitable collective physical processes can learn in Hebbian-inspired ways to recognize (chemical/mechanical) patterns without being designed to mimic a neural network element-by-element. We use examples from our work on neural computation that is latent in the nucleation of molecular structures and in bifurcations in mechanical systems. |