Applying deep learning to single-trial EEG data provides evidence for complementary theories on action control

A.AbstractThe standard approach in neuroscience research infers from the external stimulus (outside) to the brain (inside) through stimulus-evoked activity. Recently challenged by Buzsáki, he advocates the reverse; an inside-out approach inferring from the brain’s activity to the neural effects of the stimulus. If so, stimulus-evoked activity should be a hybrid of internal and external components. Providing direct evidence for this hybrid nature, we measured human intracranial stereo-electroencephalography (sEEG) to investigate how prestimulus variability, i.e., standard deviation, shapes poststimulus activity through trial-to-trial variability. We first observed greater poststimulus variability quenching in trials exhibiting high prestimulus variability. Next, we found that the relative effect of the stimulus was higher in the later (300-600ms) than the earlier (0-300ms) poststimulus period. These results were extended by our Deep Learning LSTM network models at the single trial level. The accuracy to classify single trials (prestimulus low/high) increased greatly when the models were trained and tested with real trials compared to trials that exclude the effects of the prestimulus-related ongoing dynamics (corrected trials). Lastly, we replicated our findings showing that trials with high prestimulus variability in theta and alpha bands exhibits faster reaction times. Together, our results support the inside-out approach by demonstrating that stimulus-related activity is a hybrid of two factors: 1) the effects of the external stimulus itself, and 2) the effects of the ongoing dynamics spilling over from the prestimulus period, with the second, i.e., the inside, dwarfing the influence of the first, i.e., the outside.B.Significance StatementOur findings signify a significant conceptual advance in the relationship between pre- and poststimulus dynamics in humans. These findings are important as they show that we miss an essential component - the impact of the ongoing dynamics - when restricting our analyses to the effects of the external stimulus alone. Consequently, these findings may be crucial to fully understand higher cognitive functions and their impairments, as can be seen in psychiatric illnesses. In addition, our Deep Learning LSTM models show a second conceptual advance: high classification accuracy of a single trial to its prestimulus state. Finally, our replicated results in an independent dataset and task showed that this relationship between pre- and poststimulus dynamics exists across tasks and is behaviorally relevant.

Download Full-text

A Novel Local Explainability Approach for Spectral Insight into Raw EEG-Based Deep Learning Classifiers

10.1101/2021.06.10.447983 ◽

2021 ◽

Author(s):

Charles A Ellis ◽

Robyn L Miller ◽

Vince Calhoun

Keyword(s):

Deep Learning ◽

Sleep Stage ◽

Sleep Stages ◽

Eeg Analysis ◽

Frequency Bands ◽

Learning Classifiers ◽

Eeg Data ◽

Stage Classification ◽

Sleep Stage Classification ◽

Insight Into

The frequency domain of electroencephalography (EEG) data has developed as a particularly important area of EEG analysis. EEG spectra have been analyzed with explainable machine learning and deep learning methods. However, as deep learning has developed, most studies use raw EEG data, which is not well-suited for traditional explainability methods. Several studies have introduced methods for spectral insight into classifiers trained on raw EEG data. These studies have provided global insight into the frequency bands that are generally important to a classifier but do not provide local insight into the frequency bands important for the classification of individual samples. This local explainability could be particularly helpful for EEG analysis domains like sleep stage classification that feature multiple evolving states. We present a novel local spectral explainability approach and use it to explain a convolutional neural network trained for automated sleep stage classification. We use our approach to show how the relative importance of different frequency bands varies over time and even within the same sleep stages. Furthermore, to better understand how our approach compares to existing methods, we compare a global estimate of spectral importance generated from our local results with an existing global spectral importance approach. We find that the δ band is most important for most sleep stages, though β is most important for the non-rapid eye movement 2 (NREM2) sleep stage. Additionally, θ is particularly important for identifying Awake and NREM1 samples. Our study represents the first approach developed for local spectral insight into deep learning classifiers trained on raw EEG time series.

Download Full-text