Bacteria Demonstrate Memory and Learning via Ribosome Networks, Echoing AI Systems

A team of computational biophysicists used microfluidic devices to monitor the responses of tens of thousands of individual E. coli cells as nutrient supplies were switched on and off at varying rates. The bacteria not only reacted to current nutrient levels but also incorporated their recent nutrient history, with cells from feast-and-famine conditions adapting more quickly than those from steady environments, demonstrating a form of learning.

Modeling of the internal molecular circuitry identified ribosomes as the likely memory store, with distinct fast-responding and slow-responding ribosome groups providing a multi-timescale record. This arrangement mirrors the logic of gated recurrent neural networks used in AI, where molecular “gates” regulate how much past information is retained versus overwritten. The findings reveal that chemical networks can generate memory and learning without neurons, offering a biological blueprint for energy-efficient AI and potential targets for drugs against adaptable pathogenic bacteria.