Brain-state invariant thalamo-cortical coordination revealed by non-linear encoders.

Guillaume Viejo,Adrien Peyrache,Thomas Cortier,Artur Luczak

doi:10.1371/journal.pcbi.1006041

Guillaume Viejo, Adrien Peyrache + Show 2 more

Open Access

https://doi.org/10.1371/journal.pcbi.1006041

Copy DOI

Abstract

Understanding how neurons cooperate to integrate sensory inputs and guide behavior is a fundamental problem in neuroscience. A large body of methods have been developed to study neuronal firing at the single cell and population levels, generally seeking interpretability as well as predictivity. However, these methods are usually confronted with the lack of ground-truth necessary to validate the approach. Here, using neuronal data from the head-direction (HD) system, we present evidence demonstrating how gradient boosted trees, a non-linear and supervised Machine Learning tool, can learn the relationship between behavioral parameters and neuronal responses with high accuracy by optimizing the information rate. Interestingly, and unlike other classes of Machine Learning methods, the intrinsic structure of the trees can be interpreted in relation to behavior (e.g. to recover the tuning curves) or to study how neurons cooperate with their peers in the network. We show how the method, unlike linear analysis, reveals that the coordination in thalamo-cortical circuits is qualitatively the same during wakefulness and sleep, indicating a brain-state independent feed-forward circuit. Machine Learning tools thus open new avenues for benchmarking model-based characterization of spike trains.

Highlights

Investigating how the brain operates at the neuronal level is usually addressed by the specification of neuronal responses to an experimentally measurable variable or by the quantification of the temporal coordination of neuronal ensembles [1, 2]
Since the HD signal is a well-characterized signal relative to the angular direction of the animal’s head, we compared the prediction of XGB with the output of the model-based (MB) tuning curve
We found using these methods that, in the HD system, the thalamus temporally leads the cortex during wakefulness and sleep, suggesting a bottom-up transmission of signal irrespective of the brain state

Summary

Introduction

Investigating how the brain operates at the neuronal level is usually addressed by the specification of neuronal responses to an experimentally measurable variable or by the quantification of the temporal coordination of neuronal ensembles [1, 2]. In relation to behavior or internal parameters (e.g. brain states), are evaluated by the statistical dependencies of spike trains, the most widely used method being linear cross-correlations [17]. These linear measures can be generalized to population correlation with tools such as Principal Component Analysis (PCA) [18, 19] and Independent Component Analysis [20]. Generalized linear models were used to build predictions of single spike trains as a function of the peer network activity [6] and to provide a full statistical description of spatio-temporal neuronal responses and correlations [21]. Among the large body of available tools, evaluating neuronal coupling by fitting spiking activity to Ising models has provided key insights into the nature of neuronal coordination in a population [23, 24]

Methods

Results

Conclusion