Dynamical mechanisms of astrocyte-neuron coupling for the integration of short- and long-term memory

The integration of short- and long-term memory is a fundamental feature of cognitive processing, yet the dynamical mechanisms that enable such cross-scale integration remain poorly understood. Recent studies highlight astrocytes as a central role in multi-timescale memory. Here, we explore the dynamical mechanisms of this process by developing an astrocyte-enhanced spiking neural network that integrates prefrontal working memory, motor cortical action generation, hippocampal storage and recall, and a semantic associative module. Within this framework, fast-slow astrocytic calcium dynamics jointly gate synaptic plasticity and recall activation. Obtained results demonstrate that the network maintains metastable sequential activity, supports robust cross-scale memory transformation, and exhibits dynamic transitions through nonlinear astrocytic regulation. Furthermore, by extracting hippocampal subnetworks into a stable heteroclinic circuit, we reveal how astrocyte modulation stabilizes sequential recall trajectories and facilitates the integration of short- and long-term memory. This study provides a mechanistic insight into multi-timescale memory dynamics and highlights astrocyte–neuron coupling as a key nonlinear dynamical substrate for memory stability, transfer, and contextual recall.