How Brain Signals Are Measured
Signal Acquisition
The Measurement Problem
Brain signals are extremely complex, noisy, and often weak. A computer-brain AI system must first measure neural activity with enough quality to detect useful patterns while minimizing artifacts from muscle movement, eye motion, and environmental interference.
Common Recording Methods
| Method | Strengths | Limitations |
|---|---|---|
| EEG | Noninvasive, inexpensive, portable | Low spatial resolution, noisy |
| MEG | Good temporal resolution, direct magnetic measurement | Expensive and bulky |
| fNIRS | Measures blood oxygenation, relatively wearable | Slower than electrical signals |
| ECoG | Higher fidelity than scalp EEG | Requires surgery |
| Intracortical arrays | Very detailed neural activity | Highly invasive, long-term challenges |
Resolution Trade-Off
In general, the more invasive the sensor, the higher the signal quality, but also the greater the medical risk and deployment complexity.
Typical Acquisition Pipeline
-
1
Place or implant sensors and verify signal quality.
-
2
Amplify and digitize raw neural signals.
-
3
Filter noise and remove artifacts.
-
4
Segment data into analysis windows.
-
5
Send cleaned data to decoding models.
Which recording method is usually noninvasive and portable?
EEG records scalp-level electrical activity and is commonly used because it is noninvasive and relatively portable.
Correct answer: EEG
What is one common source of noise in brain-signal recordings?
Artifacts from muscle activity, eye motion, and ambient electrical noise can contaminate signals.
Correct answer: muscle movement
Why Signal Quality Matters
No decoding model can fully compensate for poor acquisition. Good electrode placement, stable contact, proper sampling rate, and artifact rejection are foundational to any successful system.
Which statement is most accurate?
A key design principle is the trade-off between signal fidelity and invasiveness.
Correct answer: More invasive methods often provide richer signals but increase risk