Integrated Calendar

Neurons grown in vitro are in theory amenable to generate and process information as they normally do inside brains, enabling many ground-breaking applications including glucose-powered neural implants to repair cerebral functions, in vitro models for cognitive studies, and even new kinds of artificial intelligence. However, there is currently no consensus about how to harness the capabilities of neurons in culture or how to build robust and efficient neuronal devices.

Based on the properties of cultured networks that are both resilient and experimentally accessible on the one hand, and the theoretical framework of Frank Rosenblatt’s perceptron on the other hand, we propose a new architecture for neuronal devices that can bypass many limitations of previous realizations. Our devices are divided into several neuronal populations by the mean of compartmented microchips, individually acting as cohesive « switch-like » units. Between the compartments, axon tunnels of various geometries have been developed to precisely control the directions and strengths of the connections between units, thus defining the initial computational abilities of our devices.

The capacity of perceptrons to learn new computations on the fly by adjusting the strength of the connections between their units can also be implemented in such neuronal devices by exploiting synaptic plasticity, the natural ability of living neurons to change the strength of their connections when subjected to specific stimulations. Communication and learning are currently being investigated inside such devices, through a fully optical interface based on optogenetics and calcium imaging that can furthermore be implemented with off-the-shelf components.

These recent theoretical and technological developments lay a solid foundation for the study of communication, computation, plasticity and learning inside living neuronal devices, and their implementation in living neuronal computers and their applications.