{
  "id": 13155714,
  "title": "Learning Interpretable Switching Dynamics in Shared Neural-Behavioral Latent Space",
  "url": "https://urgent.news/2026/10/09/learning-interpretable-switching-dynamics-in-shared-neural-behavioral",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-09T00:00:00.000Z",
  "source": {
    "name": "bioRxiv",
    "slug": "biorxiv",
    "url": "https://www.biorxiv.org/content/10.64898/2026.10.02.756323v1?rss=1"
  },
  "original_language": "en",
  "account": "The article discusses a new modeling framework called Switching Shared Latent Dynamics (SSLD) that aims to identify the shared latent structure between neural activity and behavior, as well as the underlying dynamics that govern their evolution across behavioral states. Existing methods typically focus on either shared representation learning or dynamical modeling, but not both simultaneously. SSLD overcomes this limitation by learning a shared latent representation for neural activity and behavior while also incorporating nonlinear switching recurrent dynamics to capture the structured changes in neural representations that occur during different behavioral states. This shared space is augmented with private latent variables that account for modality-specific variability, allowing for a clearer understanding of neural activity that does not directly influence behavior. The researchers tested SSLD on four distinct experimental datasets encompassing various species, brain regions, and recording modalities, including monkey motor and premotor cortex during reaching tasks, somatosensory cortex during a bump task, widefield calcium imaging across mouse dorsal cortex during decision-making and spontaneous movements. Across all datasets, SSLD demonstrated the ability to accurately reconstruct neural and behavioral signals, identify discrete states in the dynamics that correspond to experimentally defined behavioral epochs, and isolate behaviorally-relevant neural information in the shared latent while preserving private neural variability. The study's ablation analyses further confirm the contributions of shared representation learning, behavioral supervision, and switching dynamics to SSLD's performance. Overall, SSLD provides an interpretable approach to modeling shared neural-behavioral dynamics, complementing efforts towards developing foundation models for neuroscience.",
  "summary": "Modern recording technologies enable simultaneous measurement of high-dimensional neural activity and rich behavioral variables, creating both an opportunity and a modeling challenge: identifying the shared latent structure that mediates the brain-behavior relationship and the underlying dynamics through which it evolves across behavioral states. Existing approaches focus on different parts of…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}