Emergent neuro-mimetic oscillations in engineered granular assemblies

Abstract

Brain-inspired hardware platforms break the conventional computing paradigm and promise improved power efficiency. Local features of neural function, such as spiking activity and synaptic plasticity, have been reliably mimicked in a number of hardware platforms; however, global effects such as synchronization and memory have proven more elusive. We demonstrate a non-memristive material system that emulates collective neuronal oscillations, a global feature of brain dynamics. Drawing on the biological principle that brain rhythms emerge from the interaction of opposing excitatory and inhibitory populations, we engineer an oxide nanoparticle system that exhibits spontaneous current oscillations under a constant DC bias. These oscillations display low spectral entropy, reminiscent of coherent neuronal rhythms. For small electrode separations, the current exhibits random fluctuations with higher spectral entropy, echoing the scale-dependent emergence of synchronization in neuronal populations. The system further responds to external stimuli by reconfiguring its oscillatory dynamics in the frequency domain, providing a proof-of-principle demonstration of stimulation-history-dependent memory. Kinetic Monte Carlo simulations of variable-range hopping with dynamic site occupancy reproduce the experimental dynamics and identify the scale crossover as a genuine collective synchronization rather than a statistical-averaging effect.