Inferring single-trial neural population dynamics using sequential auto-encoders

• Published in: Nature methods Vol. 15; no. 10; pp. 805 - 815
• Main Authors: Pandarinath, Chethan, O’Shea, Daniel J., Collins, Jasmine, Jozefowicz, Rafal, Stavisky, Sergey D., Kao, Jonathan C., Trautmann, Eric M., Kaufman, Matthew T., Ryu, Stephen I., Hochberg, Leigh R., Henderson, Jaimie M., Shenoy, Krishna V., Abbott, L. F., Sussillo, David
• Format: Journal Article
• Language: English
• Published: New York Nature Publishing Group US 01.10.2018; Nature Publishing Group
• Subjects: 631/114/1305; 631/378/116; 631/378/2632; Action potential; Action Potentials; Algorithms; Analysis; Animals; Bioinformatics; Biological Microscopy; Biological Techniques; Biomedical and Life Sciences; Biomedical Engineering/Biotechnology; Brain; Brain research; Coders; Dynamical systems; Dynamics; Factor analysis; Firing pattern; Firing rate; Humans; Life Sciences; Machine learning; Male; Methods; Middle Aged; Models, Neurological; Motor Cortex - physiology; Nervous system; Neural oscillations; Neurons; Neurons - physiology; Neurophysiology; Neurosciences; Population biology; Population Dynamics; Primates; Proteomics; Recording; Systems analysis; Variation
• ISSN: 1548-7091; 1548-7105; 1548-7105
• DOI: 10.1038/s41592-018-0109-9

Neuroscience is experiencing a revolution in which simultaneous recording of thousands of neurons is revealing population dynamics that are not apparent from single-neuron responses. This structure is typically extracted from data averaged across many trials, but deeper understanding requires studying phenomena detected in single trials, which is challenging due to incomplete sampling of the neural population, trial-to-trial variability, and fluctuations in action potential timing. We introduce latent factor analysis via dynamical systems, a deep learning method to infer latent dynamics from single-trial neural spiking data. When applied to a variety of macaque and human motor cortical datasets, latent factor analysis via dynamical systems accurately predicts observed behavioral variables, extracts precise firing rate estimates of neural dynamics on single trials, infers perturbations to those dynamics that correlate with behavioral choices, and combines data from non-overlapping recording sessions spanning months to improve inference of underlying dynamics. LFADS, a deep learning method for analyzing neural population activity, can extract neural dynamics from single-trial recordings, stitch separate datasets into a single model, and infer perturbations, for example, from behavioral choices to these dynamics.