Stress-Associated Alterations in Amygdala-Striatal Activity: A Multi-Level Analysis of Distributional, Dynamical, and Computational Signatures

Chronic stress is associated with alterations in neural circuit dynamics, yet the computational principles underlying these changes remain incompletely understood. Here, we reanalyzed in vivo GCaMP8s recordings from BLA-DMS and CeA-DMS projection pathways using three complementary descriptive approaches. First, Kullback-Leibler (KL) divergence was used to quantify stress-related changes in activity distributions, revealing context-dependent alterations in population-level neural responses. Second, a phenomenological second-order regression framework was employed to characterize local recovery-related dynamical features, providing a compact descriptive representation of pathway activity without implying mechanistic identification. Third, a simplified artificial neural network (ANN) was implemented as a hypothesis-generating sufficiency test. Under asymmetric optimization constraints, the model generated response profiles qualitatively resembling selected features observed experimentally. Robustness analyses indicated that similar qualitative behaviors emerged across multiple parameter initializations. Together, these results provide a multi-level descriptive characterization of stress-associated neural activity alterations across statistical, dynamical, and computational representations, while explicitly avoiding assumptions of a unified mechanistic mapping between analytical levels.