Implementation of a Minimal Recurrent Spiking Neural Network in a Solid-State Device

We study the minimal recurrent spiking neural network of a single neuron with an autaptic synapse. implement system in solid state recently introduced ultracompact (UCN) model, which is based on memristive properties thyristor. The UCN supplemented by self-synaptic, autaptic, connection, where we control feedback. Both excitatory and inhibitory cases are considered. explore systematic behavior as function intensity feedback time delay. realize tunable dynamic memory, showing graded persistent activity, short pulses allow firing rate to be controlled. finally reproduce recent experimentally observed biological autapse measured vivo, finding excellent qualitative agreement. Our work opens way into field solid-state neuroscience, accessible platform networks.