How Scientists Are Decoding Your Thoughts During Motion
For the first time, researchers can track the brain's fleeting thought patterns as we move freely through the world.
Imagine trying to record a whisper in the middle of a hurricane. For decades, this has been the monumental challenge faced by neuroscientists trying to study the human brain in its most natural state: while moving. Electroencephalography (EEG), which measures the brain's electrical activity, is notoriously sensitive to motion. Every head turn, step, or even heartbeat can drown the brain's subtle signals in a storm of electrical "noise."
Researchers have developed a powerful new method that can strip away these disruptive artifacts, enabling a fascinating technique called microstate analysis to be performed on ambulatory EEG. This advancement is opening a new window into the brain's dynamic operations as we navigate the real world, finally allowing scientists to listen to the brain's whispers even while we are in motion.
To appreciate this breakthrough, we first need to understand what scientists are trying to measure. Your brain's resting activity is not random noise; it is a highly organized, spontaneous flow of information. EEG microstates are considered the "atoms of thought"—the shortest-lasting building blocks of our conscious and unconscious cognitive processes.
Technically, they are defined as brief periods, typically lasting between 40 and 120 milliseconds, during the brain's electrical field topography across the scalp remains stable 4 . Think of them as snapshots of the brain's overall functional state. These fleeting states are thought to represent the fundamental activity of large-scale neural networks 5 .
Associated with the auditory network and phonological processing.
Linked to the visual network.
The "default mode network," active during internal thought and self-reflection.
The attention and executive control network, crucial for focused tasks.
By analyzing the timing, duration, and sequence of these microstates, scientists can infer the brain's dynamic cognitive processes without a person ever uttering a word.
The pursuit of natural brain imaging has led to the development of mobile, or "ambulatory," EEG systems. Unlike traditional setups where a person is glued to a chair in a lab, these systems allow individuals to walk, jog, and interact with their environment. But this freedom comes at a cost.
Motion artifacts are the primary obstacle. When a person moves, the electrodes shift, muscles contract, and the body generates electrical signals that are vastly larger than the brain's microvoltage-level activity. These artifacts can completely obscure the underlying neural signals, making any analysis futile.
Traditional EEG cleanup methods struggle with low signal-to-noise ratio conditions during movement.
Traditional cleanup methods, like Independent Component Analysis (ICA), often struggle in these very low signal-to-noise ratio (SNR) conditions. They can remove brain signals along with the noise or fail to clean the data sufficiently. This has severely limited the use of sophisticated analysis techniques, like microstate tracking, in real-world scenarios 1 .
The novel solution, dubbed contrast-based artifact removal, employs a sophisticated mathematical technique called Generalized Eigen Decomposition (GED). Here's a simple way to understand how it works:
The method identifies parts of the EEG signal most strongly correlated with motion artifacts by comparing high-motion periods to quiet, resting periods.
The GED algorithm acts as an intelligent filter, decomposing the EEG signal into constituent parts based on statistical power.
By leveraging the "contrast" between artifact-dominated and brain-dominated signals, the algorithm precisely removes artifacts, leaving clean brain activity.
The most remarkable feature of this approach is its efficacy even in what scientists call "ultra-low SNR conditions," where the brain's signal is almost completely buried in noise. In technical validation tests, this method achieved a stunningly high correlation of 0.93 with the true brain signal, dramatically outperforming other common methods like ICA and Artifact Subspace Reconstruction (ASR) when the noise is extreme 1 .
Behind this breakthrough are several key tools and concepts that form the modern neuroscientist's toolkit.
| Tool/Concept | Function in the Research |
|---|---|
| Ambulatory EEG | A mobile EEG system that allows brain activity to be recorded while a person is moving freely, rather than sitting still in a lab. |
| Generalized Eigen Decomposition (GED) | A powerful mathematical algorithm used to separate brain signals from motion artifacts by identifying and removing the most dominant non-brain components. |
| Global Field Power (GFP) | A measure of the total strength of the brain's electrical field at any given moment. Peaks in GFP are used to identify the most representative microstate maps 5 . |
| EEGLAB/MICROSTATELAB | A widely used, open-source software toolbox that provides a standardized pipeline for processing EEG data and performing microstate analysis 4 . |
| Semi-Simulated Data | Data created by adding real motion artifacts to clean, known brain signals. This allows researchers to rigorously test their methods because they know what the "true" brain signal should be 1 . |
To validate their new method, the researchers conducted a series of rigorous experiments, a crucial one of which can be broken down step-by-step.
Researchers recorded EEG from participants in two conditions: first at rest, and then while walking and jogging on a treadmill.
To quantitatively test the artifact removal, they used "semi-simulated" data. They took a known, clean brain signal and added real motion artifacts recorded during the treadmill sessions.
They applied the new GED-based method to this noisy, semi-simulated data, as well as to the real EEG data from the moving participants.
Using the cleaned data from the ambulatory sessions, the researchers then performed microstate analysis. They identified the canonical microstates (A, B, C, D) and calculated key parameters: their duration (how long they last), occurrence (how often they appear), and coverage (the total time they dominate) 1 5 .
The results were clear and compelling. The table below shows how the GED method excelled at recovering the true brain signal from a noisy mix, outperforming other techniques in low-SNR conditions.
| Method | Performance in Low-SNR Conditions |
|---|---|
| GED (Contrast-based) | Superior performance, high correlation (0.93) with ground truth. |
| Artifact Subspace Reconstruction (ASR) | Less effective than GED in very low SNR regimes. |
| Independent Component Analysis (ICA) | Struggled to perform well in very low SNR regimes. |
| Source: Adapted from 1 | |
When applied to the real walking and jogging data, the GED method successfully cleaned the signals, increasing the number of identifiable brain components by over 10 times in some cases. This clean data was of high enough quality to reveal how brain dynamics actually change with motion.
| Microstate | Associated Network | Key Change During Motion |
|---|---|---|
| A | Auditory / Phonological | Increased duration, occurrence, and time coverage. |
| B | Visual | Increased duration. |
| C | Default Mode | No significant change highlighted in this study. |
| D | Attention / Executive | Decreased occurrence and time coverage. |
| Source: Adapted from findings in 1 | ||
The most exciting outcome was what the cleaned data revealed about brain function. The changes in microstate dynamics paint a coherent picture of the brain in motion. The increase in microstate A (linked to auditory processing) and B (visual processing) suggests a heightened alertness and increased sensory processing as the brain navigates a complex environment. Meanwhile, the decrease in microstate D, associated with focused attention, might indicate a shift away from deep internal reflection and toward a more reactive, externally-focused state 1 .
This breakthrough in contrast-based artifact removal is more than just a technical feat; it represents a paradigm shift. By enabling microstate analysis during free movement, it unlocks the potential to study the human brain in its most authentic context.
Imagine analyzing the neural dynamics of a conversation between two people both walking and talking.
It could lead to new biomarkers for neurological and psychiatric conditions like Parkinson's, Alzheimer's, or schizophrenia by observing brain network dynamics during natural behavior, not just in a sterile clinic 5 .
Researchers could track how the brain's dynamic networks mature in children as they play and explore their world.
The ability to unshackle the brain from the lab chair and decode its fundamental states as we live our lives marks the dawn of a new era in human neuroscience. We are no longer just listening to the brain; we are starting to hear its story as it was meant to be told—in motion.