Unsupervised discovery of demixed, low-dimensional neural dynamics across multiple timescales through tensor components analysis

Abstract

Perceptions, thoughts and actions unfold over millisecond timescales, while learned behaviors can require many 2 days to mature. While recent experimental advances enable large-scale and long-term neural recordings with high 3 temporal fidelity, it remains a formidable challenge to extract unbiased and interpretable descriptions of how rapid 4 single-trial circuit dynamics change slowly over many trials to mediate learning. We demonstrate a simple tensor 5 components analysis (TCA) can meet this challenge by extracting three interconnected low dimensional descriptions 6 of neural data: neuron factors, reflecting cell assemblies; temporal factors, reflecting rapid circuit dynamics mediating 7 perceptions, thoughts, and actions within each trial; and trial factors, describing both long-term learning and trial-8 to-trial changes in cognitive state. We demonstrate the broad applicability of TCA by revealing insights into diverse 9 datasets derived from artificial neural networks, large-scale calcium imaging of rodent prefrontal cortex during maze 10 navigation, and multielectrode recordings of macaque motor cortex during brain machine interface learning.

ERAASR: an algorithm for removing electrical stimulation artifacts from multielectrode array recordings

“…This problem of finding shared structure naturally lends itself to tensor decomposition (Kolda and Bader, 2009), which could identify shared structure jointly over each of these axes and potentially improve the artifact estimation. Tensors naturally arise in neuroscientific data collection contexts, and decomposition methods are becoming increasingly useful for identifying population structure (Seely et al, 2016; Elsayed and Cunningham, 2017; Williams et al, 2017).…”

Spiking neural network models of sound localisation via a massively collaborative process

“…As an alternative approach, we also used Tensor Component Analysis (TCA) (Williams et al, 2018) to explore the spiking activity of this model, and to compare it across multiple trained networks (Section 7.b).…”

Understanding Sensory Information Processing Through Simultaneous Multi-area Population Recordings

“…Unsupervised approaches will operate on the data blind to these experimental manipulations or outcomes, and the components they extract may not isolate the impact of experimental variables of interest (Kobak et al, 2016). To address this shortcoming, a layer of supervision can be added to isolate experimental variables, e.g., hierarchical decomposition (Repucci et al, 2001; Maddess et al, 2006), demixed PCA (Kobak et al, 2016), and tensor component analysis (Williams et al, 2017). This means that the recovered components are those that best explain individual and paired factors of interest (Brendel et al, 2011; Kobak et al, 2016).…”