Based on brain‐inspired computing frameworks, neuromorphic systems implement large‐scale neural networks in hardware. Although rapid advances have been made in the development of artificial neurons and synapses in recent years, further research is beyond these individual components and focuses on neuronal circuit motifs with specialized excitatory–inhibitory (E–I) connectivity patterns. In this study, we demonstrate a core processor that can be used to construct commonly used neuronal circuits. The neuron, featuring an ultracompact physical configuration, integrates a volatile threshold switch with a gate‐modulated two‐dimensional (2D) MoS₂ field‐effect channel to process complex E–I spatiotemporal spiking signals. Consequently, basic neuronal circuits are constructed for biorealistic neuromorphic computing. For practical applications, an algorithm‐hardware co‐design is implemented in a gate‐controlled spiking neural network with substantial performance improvement in human speech separation.