image: Visible light irradiation is analogous to food, UV irradiation is analogous to the bell ringing, and the green fluorescence is analogous to the salivation of the dog.
Credit: ©Science China Press
A research team led by Professor Han Zhang at Shenzhen University has pioneered a novel optical neural network that learns like a living organism—without relying on traditional computing algorithms. Published recently in National Science Review, the work draws direct inspiration from Ivan Pavlov’s century-old “dog and bell” experiment.
Instead of using energy-intensive backpropagation, the team engineered a dual-color photoresist material that physically “learns” through associative light exposure. When ultraviolet (UV) light is followed by visible (green) light, the resin undergoes a permanent chemical change, enabling it to later emit green fluorescence upon UV stimulation alone—mimicking a conditioned reflex.
This mechanism allows the optical network to be trained directly by light patterns. In demonstrations, it successfully recognized letters ‘N’, ‘V’, and ‘Z’; simulations further showed its capability in handwritten digit recognition. Crucially, the system eliminates the need for pre-computed weights or electronic processing, offering a “top-down,” in-situ training approach.
The technology promises ultra-low-cost, passive, and robust photonic AI hardware ideal for edge computing applications—from smart sensors to real-time industrial monitoring—ushering in a new paradigm where materials themselves embody intelligence.