Inferring Multi-Timescale Neural Dynamics with Switching Linear Dynamical Systems

2026-10-01

AI Summary

This study addresses the challenge of accurately extracting multi-timescale dynamical features from high-dimensional neural data using traditional methods. To this end, we propose the MTS-SLDS framework based on switching linear dynamical systems. By integrating multi-lag moment initialization with a state-conditional Laplace EM inference algorithm, the method explicitly models multiple timescales to prevent statistical mixing of dynamics across states and directly extracts characteristic timescales via eigenvalue analysis of the transition matrix. Experiments on both synthetic and real Gaussian and Poisson spiking neural data demonstrate that the proposed framework accurately recovers latent timescales and state-switching structures. These results establish MTS-SLDS as an effective tool for dissecting complex neural dynamics.

Abstract

Neural activity often exhibits multiple timescales that can vary with behavioral states and task conditions. Identifying these timescales from neural recordings is important for better understanding neural computation and function. However, traditional approaches based on autocorrelation fitting are difficult to scale to high-dimensional population recordings and can become unreliable when neural dynamics change with behavior. State-space models have been a powerful framework for modeling high-dimensional neural population activity through latent dynamical systems, but standard formulations and inference methods do not explicitly account for multiple timescales and therefore do not guarantee accurate recovery of the underlying temporal structure. Motivated by these questions, we introduce the Multi-Timescale Switching Linear Dynamical System (MTS-SLDS), a framework for identifying regime-specific latent timescales from continuous or spiking neural observations. MTS-SLDS combines a multi-lag moment initialization, which captures temporal structure across multiple observation lags, with \textit{regime-conditioned} Laplace-EM inference, which reduces mixing of dynamical statistics across uncertain regimes. Characteristic timescales can then be extracted directly from the eigenvalues of the learned latent transition matrices. In synthetic and neural experiments with Gaussian and Poisson spike observations, MTS-SLDS accurately recovers timescales and switching structure over multiple datasets.

Problem

• multi-timescale neural dynamics
• state-space models
• switching linear dynamical systems
• neural population recordings
• timescale identification

Innovation

• Multi-Timescale Switching Linear Dynamical System
• Laplace-EM inference
• multi-lag moment initialization
• latent timescales
• state-space models