BB.lv: Artificial brain dendrites created

The new system was implemented using a computational model of multigate silicon nanowire transistors with ion-doped films obtained by the sol-gel method. The film imitates dendritic branches, allowing the doped ions to move like ions in neuronal processes. In this case, the current inside the transistor is modulated similar to how changes in the action potential (neuronal signal) occur in the dendrite membrane.

The device, called a dendristor, exhibits nonlinear properties and direction selectivity. Most neuromorphic devices presented so far have mimicked the synaptic processes associated with learning and reproduced the generation of action potentials. These studies treated dendrites as simple transmission lines, thereby ignoring the functions associated with their unique morphology.

An artificial system with silent synapses was also presented, where synaptic inputs are activated only when a certain threshold is reached, which improves the system’s ability to discriminate the directions of moving stimuli, as occurs in the retina and visual cortex.

The results of the study may open new avenues for the design of neuromorphic systems based on semiconductor devices. In particular, the brain-inspired design proposed by the scientists may contribute to the development of new devices and AI tools that consume less energy.