Toward chips that mimic the brain: they learn and forget like neurons

An international group of researchers led by the Institute of Nanotechnology of the CNR in Lecce has developed chips capable of mimicking not only the learning process of the human brain but also its mechanism of forgetting. The result, described in the journal Communications Materials, paves the way for ultra-fast computers with extremely low energy consumption.

The new chips were achieved thanks to a new class of materials capable of simultaneously transporting electrical charges and biological signals. As explained by the research coordinator Pierluigi Cossari of CNR-Nanotec, being able to regulate forgetting, that is, the ability to 'forget' information that is no longer necessary, is essential for optimizing resources, just as happens in biological neurons.

Research on neuromorphic computers, which mimic the architecture and functioning of the human brain, is a frontier field in technology. The goal is to overcome the limitations of traditional computers: although they are extremely fast at processing, they consume enormous amounts of energy. The human brain, on the other hand, performs equally complex operations with much lower energy consumption, thanks to its structure based on neurons and synapses.

The path to achieving this result has seen researchers draw inspiration from biological mechanisms to create 'intelligent' artificial synapses. The collaboration between various institutes of the CNR (crystallography in Bari, microelectronics in Lecce, nanostructured materials in Rome), together with the University of Calabria and the Italian Institute of Technology in Rome, has enabled the development of innovative organic materials. These materials have the unique characteristic of integrating electronic and biological functionalities, finally allowing the replication not only of the learning process but also the physiological process of forgetting, a crucial step toward truly efficient and adaptive computing systems.