Inside Muse S Athena: Unpacking the EEG + fNIRS Dual-Sensor Architecture 

The Muse S headband has become one of the more common consumer EEG devices, partly because it actually delivers usable signal, and partly because the meditation and sleep tracking features appeal to people outside the research space. The newest version, the Muse S Athena adds something different: fNIRS sensors alongside the EEG. This is worth examining not because it's revolutionary, but because it represents a real engineering shift in how consumer neurotech approaches brain monitoring. 

What the Hardware Actually Does

Let's start with basics. The Muse S Gen 2 uses four dry electrode channels positioned on the forehead, plus two auxiliary channels for reference. The amplifier is 12-bit resolution, sampling at 64 Hz. That's sufficient for detecting broad brainwave patterns, the differences between relaxed alpha waves and focused beta waves are fairly distinct even at 64 Hz. Where it gets thin is artifact rejection. At 64 Hz with 12-bit resolution, you're trading temporal and amplitude precision for simplicity and battery life

The Muse S Athena bumps the EEG to fourteen channels with 14-bit resolution. On paper, this doubles your signal precision. In practice, it means slightly less quantization noise when the signal is small, which matters if you're trying to detect subtle changes in brain activity. Whether this actually improves real-world neurofeedback performance is harder to say I haven't seen a head-to-head comparison showing statistically significant differences in training outcomes between the two versions.

The hardware change is worth noting: silver-thread fabric electrodes instead of silver ink. Silver oxidizes over time, which degrades contact. The fabric version isn't a complete solution to this problem, it still oxidizes but it maintains contact better and doesn't require paste or gel. That's a practical engineering decision that makes overnight wear actually feasible.

The fNIRS Addition: What Changes, and What Doesn't

fNIRS (functional near-infrared spectroscopy) measures hemoglobin oxygenation by shining near-infrared light into tissue and measuring absorption patterns. The physics is straightforward: oxygenated and deoxygenated hemoglobin have different absorption profiles. The engineering challenge is filtering out motion artifacts, blood pressure changes, and superficial skin blood flow to isolate brain activity.

The Athena's fNIRS setup uses wavelengths between 600-1000 nm, which penetrate roughly 1-2 cm into tissue deep enough to reach cortical regions in the prefrontal cortex, but not deep enough to see much of the temporal lobe or anything subcortical. This is a fundamental limitation, not a solvable one. fNIRS is cortical surface imaging.

The claimed advantage of combining EEG and fNIRS is complementary resolution. EEG has good temporal resolution (you see millisecond-level brainwave oscillations) but poor spatial resolution (you can't pinpoint where activity is happening with much precision). fNIRS has worse temporal resolution but better spatial localization. Theoretically, combining them gives you both temporal and spatial information.

The research support for this is legitimate but not overwhelming. Studies show that EEG+fNIRS fusion improves classification accuracy for mental states compared to either modality alone. One study on ADHD diagnosis achieved 93% accuracy with combined EEG+fNIRS versus 79% with EEG alone using machine learning approaches. That's a meaningful difference. But most of these studies involve controlled lab settings with careful signal preprocessing. Consumer devices operating in homes with pets, WiFi interference, and variable electrode contact are noisier environments.

What the Device Actually Does in Practice

Meditation training: The real-time feedback during meditation is the clearest use case. As you meditate, increased mind-wandering produces beta and gamma waves, which trigger audio cues (typically birds or storms) to alert you. Refocusing quiets the signal, audio recedes, and you get a reinforcing feedback loop. This is basic operant conditioning. It works. Studies confirm it improves attention compared to guided meditation without feedback. The question is whether it's better enough to justify wearing a headband compared to just practicing meditation normally. That's subjective.

Sleep tracking: The Muse S claims 86% accuracy in sleep stage classification compared to polysomnography. This needs context. Other wearables claim 80-85% accuracy too. Clinical PSG isn't perfect either raters can disagree on stage classification, and the "gold standard" criterion changes as research evolves. Also, 86% overall accuracy masks variation between stages. Deep sleep classification is often less accurate than REM or light sleep detection, which matters if someone has specific sleep problems. Users report discrepancies, the device sometimes underestimates deep sleep compared to their subjective sleep quality.

The Digital Sleeping Pill (DSP) feature using real-time EEG to detect sleep onset and adjust audio is technically interesting but hard to evaluate objectively. Subjective reports are positive, but individual variation is huge. Some people find it genuinely helpful; others find wearing the headband during sleep disruptive. 

Focus training with fNIRS: 

The Athena's owl game is new. As your prefrontal cortex blood oxygenation increases, the owl climbs; as mental effort drops, it falls. Eyes-open training matters because it mimics real work, unlike closed-eye meditation.

But does fNIRS-based training actually work better than EEG-based? Research shows fNIRS feedback can train brain activation people learn to increase prefrontal oxygenation when they see real feedback but can't when feedback is fake. A 2024 study found that eight sessions of fNIRS training produced measurable changes in how the prefrontal cortex connects to deeper brain regions involved in learning. However, roughly 27% of people showed consistent results across all sessions, while others were less responsive, suggesting significant individual variation.

The gap: no one has directly compared fNIRS training versus EEG training using the same protocol. The prefrontal cortex's blood oxygenation is theoretically more specific to focus than general brainwave patterns, but that doesn't automatically mean it trains better in practice. Studies on neurofeedback for attention show modest improvements overall, with consistency mattering more than which brain signal you monitor. The Athena just launched; actual evidence comparing it to EEG-only devices won't exist for another year.

The Practical Limitations

Setup is finicky. Poor electrode contact kills your signal, and the app guides positioning but you need to sit still and position the headband correctly. Any movement creates noise spikes. The app flags bad signal quality, but knowing about a problem doesn't solve it, you still have to restart and reposition.

fNIRS only sees surface activity. It measures oxygenation in the prefrontal cortex (front of your brain) but can't reach deeper regions or the sides. If your task uses other brain areas, fNIRS misses it. This is a fundamental limit of the technology, not just this device.

The 14-bit EEG resolution is okay but basic. Clinical systems use 24-bit; research devices like the Emotiv EPOC use 16-bit; open-source systems go to 24-bit. The Athena's resolution creates more noise with small signals. For casual use this is fine, but if you're serious about signal analysis, you'll hit the ceiling quickly.

Constant Bluetooth means constant wireless connection. Some people worry about this; some don't. Recording to the device directly would reduce exposure but would drain battery faster and add complexity, so Muse went with streaming.

Advanced features require a subscription. Without paying $50/year or $12.99/month, you lose Alpha Peak tracking, detailed sleep analysis, 500+ meditation sessions, and audio integration. For a $475 device, paying extra for core features feels extractive. Though to be fair, new content and algorithm updates have real costs.

Positioning in the Landscape

The Muse S Athena sits in an interesting price/capability space. At $475 (without subscription), it's significantly cheaper than clinical EEG systems ($3,000-15,000+), considerably more expensive than consumer fitness wearables ($200-400), and in the same ballpark as serious DIY research devices like the Emotiv EPOC ($800).

The EPOC offers more channels (14 channels), higher bit resolution (16-bit), and doesn't require a subscription. But you're paying for engineering flexibility and data access rather than polished consumer features. The Muse has better meditation/sleep features and app quality; the EPOC is better for research or serious DIY signal processing.

Clinical neurofeedback systems like Sens.ai cost $1,500+ and involve professional protocols. They're different products for different use cases.

What Actually Matters

From a technical standpoint, the most interesting thing about the Athena isn't that EEG+fNIRS is combined that's been done in research labs for years. What's interesting is the implementation: fitting both into a wearable form factor, streaming both modalities via Bluetooth without overwhelming battery life, and developing biofeedback algorithms that leverage both signals.​

Whether the dual-modality approach actually improves training outcomes compared to EEG-only is still an open question. The device hasn't been out long enough for robust third-party validation. The device works; users report benefits; the underlying physiology is sound. But marketing claims about cognitive enhancement aren't the same as controlled evidence.

If you're considering buying one, the most honest summary is: you're getting an EEG-based meditation trainer with decent signal quality, a sleep tracker with reasonable accuracy, and a new fNIRS channel for real-time focus feedback. The neurofeedback model is backed by research, but it's not a clinical tool. If daily meditation practice, sleep tracking, and real-time focus feedback appeal to you, it's probably worth it. If you're hoping it will substitute for treating actual cognitive or sleep disorders, talk to a clinician first.

References

1. Bitbrain. (2024, April 23). Fusion of fNIRS and EEG: a step further in brain research. https://bitbrain.com

2. Choose Muse. (2025, March 20). Athena vs. Muse S G2: The biggest brain tech upgrade yet. https://choosemuse.com/blogs/news/athena-vs-muse-s-g2-the-biggest-brain-tech-upgrade-yet

3. Choose Muse. (2025, April 1). Can Muse S help long COVID symptoms? A 180-day study reveals the data. https://choosemuse.com/blogs/news/can-muse-s-help-long-covid-symptoms-a-180-day-study-reveals-the-data

4. Cybernews. (2025, July 9). Muse S Athena review 2025. https://cybernews.com/health-tech/muse-s-review/

5. DIY Genius. (2025, April 23). How the Muse S Athena works for EEG & fNIRS neurofeedback. https://diygenius.com/how-the-muse-s-athena-eeg-and-fnirs-neurofeedback-device-works/

6. Good Gear. (2025, June 22). A therapist tries the Muse at-home biofeedback headband. https://www.goodgear.com/muse-headband-review/

7. Kummer, M. (2025, June 16). Muse S Athena review: A brain-sensing headband that changed my life. https://michaelkummer.com/muse-s-athena-review

8. Outliyr. (2025, August 24). Muse Athena review (Gen 3): Can it hack your brain, sleep and focus?. https://www.outliyr.com/muse-athena-review-gen-3