How Scientists Learned to Isolate the Brain's True Message
The brain's whispers are nearly drowned out by the roar of blood flow, but a clever technique finally separates the two.
Imagine you're trying to listen to a faint symphony from outside a concert hall. Your ear is pressed against the wall, but all you can clearly hear is the rain drumming on the roof and the rustling of leaves in the trees around you. These distracting sounds aren't part of the music, but they're overpowering the delicate melodies you're trying to hear. This is precisely the challenge neuroscientists faced for years when using a popular brain imaging technique called functional near-infrared spectroscopy (fNIRS).
fNIRS is a remarkable non-invasive technology that shines harmless near-infrared light through the scalp and skull to measure brain activity by tracking changes in blood oxygenation 9 . It's portable, relatively inexpensive, and allows people to move freely—making it ideal for studying brain function during real-world activities like walking, playing instruments, or even while patients undergo rehabilitation 2 .
However, this powerful tool had a hidden flaw: it couldn't reliably distinguish between actual brain activity and blood flow changes happening in the scalp itself. That is, until a pivotal 2017 study revealed how these "systemic interferences from the superficial layer" were misleading scientists, particularly during brief motor tasks, and offered an elegant solution to the problem 3 . This discovery not only resolved a methodological headache but opened new possibilities for accurately decoding the brain's complex language.
Functional near-infrared spectroscopy is a non-invasive brain imaging method that uses light to measure changes in blood oxygenation related to neural activity.
To understand why this discovery mattered, we first need to understand how fNIRS works and where it was being fooled.
fNIRS operates on a simple but ingenious principle: active brain regions demand more oxygen, triggering increased blood flow to those areas. This creates a predictable signature change in the concentration of oxygenated hemoglobin (HbO) and deoxygenated hemoglobin (HbR) in the blood .
Near-infrared light can penetrate biological tissues, and crucially, hemoglobin absorbs this light differently depending on whether it's carrying oxygen. By emitting light at specific wavelengths and measuring how much reaches detectors placed on the scalp, scientists can calculate these hemoglobin concentration changes, thereby inferring which brain areas are active 9 .
Light absorption changes indicate brain activity
The fundamental challenge arises because the light measured by fNIRS detectors has traveled through multiple layers: the skin, scalp, skull, and finally, the brain tissue. The signal contains a mixture of information from all these layers. Task-evoked systemic artifacts—blood flow changes in the scalp itself—were found to be strong enough to mask the cerebral activation signals scientists sought to detect 3 8 .
Think of it like trying to measure the temperature deep in a lake while being unable to separate your reading from the warming effect of the sun on the surface. For years, researchers wondered why brain activation signals during certain tasks seemed inconsistent or sometimes even indicated activation where there shouldn't be any.
This problem was particularly acute during event-related motor tasks—brief, movement-based activities like quickly squeezing a ball—because the scalp itself has its own robust blood supply that responds dynamically to physical exertion, stress, or even just the anticipation of movement 3 .
Near-infrared light emitted into scalp
First layer with its own blood flow changes
Bone layer with minimal interference
Target region with neural activity signals
Measures light that has traveled through all layers
The 2017 study, published in the Journal of Biomedical Optics, marked a significant turning point by systematically investigating how these superficial interferences influenced event-related motor tasks and proposing a practical solution 3 .
The researchers designed an elegant experiment centered around a simple brisk squeezing movement. Participants performed this motor task while wearing an fNIRS cap, but with a crucial modification to the standard setup.
The key innovation was the use of short source-detector separation channels (approximately 15 mm). At this short distance, the light predominantly samples the superficial layers (skin and scalp) without reaching the brain in significant quantities. These short-distance channels therefore act as a dedicated "interference monitor," capturing the hemodynamic noise originating from the scalp 3 .
The experimental process unfolded as follows:
Participants engaged in event-related motor tasks as well as traditional block-design motor tasks.
fNIRS data was simultaneously collected from both short-distance and long-distance channels.
Using principal component analysis (PCA), researchers estimated superficial-tissue hemodynamics.
Estimated superficial interference was subtracted from long-distance signals.
Short Distance
~15mm
Long Distance
~30mm
Short-distance channels primarily capture signals from the scalp, while long-distance channels capture mixed signals from both scalp and brain.
The results were both startling and illuminating, revealing two critical patterns that would change how fNIRS data is processed.
The researchers discovered that the task-evoked hemodynamic responses in the superficial layer—the "noise"—looked remarkably similar to the canonical cerebral hemodynamic response expected from actual brain activation. They exhibited the same pattern of a rapid increase in oxygenated hemoglobin followed by a gradual return to baseline 3 .
This superficial interference occurred regardless of the task design, appearing in both the block-design and event-related motor tasks. This demonstrated that the problem was fundamental and not limited to a specific experimental approach 3 .
Most significantly, when the researchers applied their correction method—removing the estimated superficial interference using data from the short-distance channels—the estimation of event-related cerebral hemodynamics improved dramatically. The true brain signal emerged from the noise, providing a more accurate picture of what was actually happening in the brain during these brief motor tasks 3 .
Signal clarity after correction
| Characteristic | Superficial Layer Response | Cerebral Brain Response |
|---|---|---|
| HbO Change | Transient increase | Transient increase |
| Time Course | Similar to canonical model | Canonical model |
| Origin | Scalp blood flow | Cortical neural activity |
| Influence on fNIRS | Can mask true brain signals | Target of measurement |
| Task Design Dependence | Occurs in both designs | Occurs in both designs |
| Signal Metric | Before Correction | After Correction |
|---|---|---|
| Cerebral activation identification | Obscured by systemic artifacts | Improved clarity |
| False-positive risk | Higher | Reduced |
| Resemblance to canonical HRF | Distorted | More accurate |
| Reliability for event-related tasks | Lower | Higher |
To conduct this type of sophisticated fNIRS research, scientists rely on specialized tools and methodological approaches.
| Research Component | Function & Purpose |
|---|---|
| Short Source-Detector Separation Channels | Samples hemodynamic changes primarily from superficial layers (skin/scalp) to isolate interference. |
| Principal Component Analysis (PCA) | Statistical technique to identify and separate systemic interference patterns from cerebral signals. |
| Time-Domain fNIRS | Advanced method measuring photon time-of-flight to provide depth-resolution and separate layer-specific signals 8 . |
| Event-Related Motor Paradigm | Experimental design using brief, discrete movements to study transient brain activation. |
| Two-Layered Phantom Models | Laboratory validation systems simulating scalp and brain layers to test signal separation algorithms 8 . |
Statistical method that identifies patterns in data and separates signal components based on variance.
Laboratory systems that simulate human tissue properties to validate measurement techniques.
Advanced version of fNIRS that measures photon travel time for better depth resolution.
The ramifications of this research extend far beyond technical methodology, influencing how we study the brain in both health and disease.
The findings sparked what might be called a "signal purity revolution" in fNIRS research. They provided conclusive evidence that without proper signal separation, fNIRS data—particularly from studies involving movement—could be significantly contaminated, potentially leading to false positives or misleading conclusions about brain activation patterns.
This has led to new standards in fNIRS experimental design, with many researchers now routinely incorporating short-distance channels as a necessary control for superficial hemodynamic changes. The methodology has proven especially valuable in studies involving:
This improved ability to capture clean signals during movement has accelerated fNIRS applications in real-world scenarios. Researchers can now more confidently study brain activity during walking, balance control, and upper limb movements . The technology's portability, combined with robust signal processing, makes it possible to take neuroimaging out of the laboratory and into environments where natural movement occurs.
Furthermore, this methodological advancement has strengthened fNIRS as an ideal partner for combined neurostimulation and neuroimaging approaches. Since fNIRS is relatively immune to the electromagnetic interference created by techniques like transcranial magnetic stimulation (TMS) or transcranial direct current stimulation (tDCS), it can simultaneously monitor brain responses to these neuromodulation therapies 1 5 9 .
This synergy is particularly promising for developing closed-loop neurorehabilitation systems where brain activity is monitored in real-time during therapeutic stimulation, allowing for personalized, precisely-timed interventions that could significantly improve outcomes for patients with neurological conditions 9 .
Monitoring recovery after stroke or injury
Tracking brain development in children
Studying brain activity during athletic performance
Developing more responsive neural interfaces
The journey to solve the problem of superficial interference in fNIRS represents more than just a technical fix—it exemplifies the iterative, self-correcting nature of science.
What began as a persistent nuisance obscuring brain signals has become a solved problem with elegant methodologies that now strengthen the reliability of an entire field.
Thanks to this crucial work, scientists can now listen more clearly than ever to the brain's complex hemodynamic symphony, distinguishing the true melody of cortical activation from the distracting rhythms of scalp blood flow. As fNIRS continues to evolve, finding applications in everything from brain-computer interfaces to clinical neuromodulation therapies, this foundational advancement ensures that the stories we tell about brain function are built on increasingly solid, unambiguous evidence.
The next time you see someone wearing what looks like a colorful swim cap studded with sensors during a neuroscience study, remember that behind each of those sensors lies a sophisticated understanding of how to separate the brain's true message from the noisy world it inhabits.