Discovering Precise Temporal Patterns in Large-Scale Neural Recordings through Robust and Interpretable Time Warping

• Published in: Neuron (Cambridge, Mass.) Mass.), 2020-01, Vol.105 (2), p.246-259.e8
• Main Authors: Williams, Alex H., Poole, Ben, Maheswaranathan, Niru, Dhawale, Ashesh K., Fisher, Tucker, Wilson, Christopher D., Brann, David H., Trautmann, Eric M., Ryu, Stephen, Shusterman, Roman, Rinberg, Dmitry, Ölveczky, Bence P., Shenoy, Krishna V., Ganguli, Surya
• Format: Article
• Language: English
• Subjects: Action Potentials - physiology; Amyloid beta-Protein Precursor - genetics; Animals; Behavior; Cognition; Cortex (motor); Cortex (temporal); Exploratory Data Analysis; Firing pattern; Gene Knock-In Techniques; Macaca mulatta; Male; Mice; Mice, Transgenic; Microinjections; Motor Cortex - physiology; Neural Oscillations; Neurons; Neurons - physiology; Noise; Olfaction; Oscillations; Pattern Recognition, Automated - methods; Peptide Fragments - genetics; Population Dynamics; Primary Cell Culture; Proteins - genetics; Rats; Signal processing; Spike Train Analysis; Statistical methods; Statistics; Time Factors; Time series; Time Warping; Unsupervised Learning
• ISSN: 0896-6273; 1097-4199
• DOI: 10.1016/j.neuron.2019.10.020

Though the temporal precision of neural computation has been studied intensively, a data-driven determination of this precision remains a fundamental challenge. Reproducible spike patterns may be obscured on single trials by uncontrolled temporal variability in behavior and cognition and may not be time locked to measurable signatures in behavior or local field potentials (LFP). To overcome these challenges, we describe a general-purpose time warping framework that reveals precise spike-time patterns in an unsupervised manner, even when these patterns are decoupled from behavior or are temporally stretched across single trials. We demonstrate this method across diverse systems: cued reaching in nonhuman primates, motor sequence production in rats, and olfaction in mice. This approach flexibly uncovers diverse dynamical firing patterns, including pulsatile responses to behavioral events, LFP-aligned oscillatory spiking, and even unanticipated patterns, such as 7 Hz oscillations in rat motor cortex that are not time locked to measured behaviors or LFP. [Display omitted] •Trial-to-trial variability in spike timing is correlated across co-recorded neurons•Unsupervised time warping reveals precise spike patterns from neural data alone•Simple and interpretable warping functions (piecewise linear) are often sufficient•Spike time precision may be systematically underestimated without warping The timing of neural dynamics can be highly variable across trials due to uncontrolled behavioral variability or unobserved cognitive states. Williams et al. describe an interpretable statistical model to control for these misalignments and use this approach to uncover fine-scale temporal structure that is imperceptible in raw data.