Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
We introduce the Multi-Timescale Switching Linear Dynamical System (MTS-SLDS), a novel framework designed to infer regime-specific latent timescales from high-dimensional neural population recordings, addressing limitations of traditional autocorrelation fitting and standard state-space models. By combining multi-lag moment initialization with regime-conditioned Laplace-EM inference, MTS-SLDS effectively disentangles dynamical statistics across behavioral states and accurately recovers characteristic timescales from learned latent transition matrices. Experiments on synthetic and real neural data demonstrate the framework's ability to reliably identify both temporal structures and switching dynamics.
Neural dynamics aren't monolithic; a new model precisely disentangles and quantifies multiple, behavior-dependent timescales from population recordings.
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.