Researchers uncover how silent neurons regenerate activity via synaptic memory

A new paper from physicists Mozhgan Khanjanianpak and Alireza Valiadeh introduces a minimal model of neuronal networks that reveals how activity can spontaneously regenerate after complete neuronal silence. The key insight is that transient synaptic memory — short-lived changes in synaptic strength — carries enough information to determine whether the network will fire again. By analyzing the synaptic state at the first moment of silence, the authors define a metric called Latent Excitatory Recruitment (LER) capacity, which counts the number of fresh excitatory neurons available for recruitment. This single number predicts with near-perfect accuracy whether the network will undergo one or two activity cycles, without needing to simulate further.

The findings challenge the traditional view that short-term memory resides solely in ongoing neuronal firing. Instead, the residual synaptic configuration itself encodes future potential. The researchers show that in an otherwise homogeneous network, transient synaptic memory alone can produce diverse dynamical outcomes. This work has direct implications for understanding biological memory consolidation, but also for designing more efficient artificial neural networks — especially neuromorphic systems that operate on sparse or intermittent activity. The snapshot-based framework could eventually be used to control network evolution, making it a practical tool for computational neuroscience and next-generation AI that mimics biological efficiency.