"Neurons and electricity: where the EEG signal comes from"
- EEG
- EEG Beginner
- Neuroscience
- Data Engineering
- Signal Processing
Part 2 of 28 · Beginner · EEG from zero: the series
An EEG electrode can't hear one brain cell. It hears the sum of millions of cells that line up and work in step. This lesson explains which currents EEG picks up, and why they only add up when the cells are in sync.
The one-sentence idea
Scalp EEG is the sum of slow currents from many aligned neurons, and it only grows large when they work in step.
Explain it simply
Your brain has about 86 billion neurons (nerve cells). About 16 billion of them sit in the cortex, the brain's outer layer. Neurons pass messages using small electric currents.
Now picture a stadium again. One person singing is lost in the noise. When 50,000 people sing the same chant at the same time, you hear it from the car park.
The brain works the same way. When many neurons are active at different moments, their currents cancel out. When they are active together, the currents add up into a signal an electrode can measure.
If you work with data, you already know this rule. Add up many signals that are out of step, and they mostly cancel. Add up signals that are in step, and they grow with every one you add.
Go one level deeper
Neurons make two kinds of electrical events:
| Event | What it is | How long it lasts | Seen by scalp EEG? |
|---|---|---|---|
| Action potential | The "spike" a neuron fires to send a message down its fibre | About 1 ms | Hardly ever |
| Postsynaptic potential | The response in the receiving neuron, where the message arrives | Tens of milliseconds | Yes, when many happen together |
Why the slow events win. An action potential is so short that spikes from different cells rarely overlap in time.
Postsynaptic potentials last much longer. So thousands of them overlap and add up.
Why the shape of the cells matters. The main signal comes from pyramidal neurons. Each one has a long branch, the apical dendrite, that points up towards the surface of the brain.
These cells stand side by side like trees in a forest.
When one of them is active, current flows along that branch. That makes a tiny dipole: a pair of opposite charges, like the two ends of a battery.
Because the cells point the same way, their dipoles add up. Cells whose branches point in all directions cancel each other out.
How much cortex it takes. A study comparing scalp and in-brain recordings found that scalp spikes usually came from 10 to 20 cm² of cortex active together.
That is a patch about 3 to 4.5 cm across. Smaller patches mostly went unseen.
Which way the dipole points matters too. EEG picks up best the dipoles that point straight out towards the scalp. These sit mostly on the tops of the brain's folds.
MEG picks up best the ones that lie sideways, in the grooves between folds. That is one reason the two methods complement each other.
Visual: many neurons, one signal
Watch the trace at the bottom. While the cells are out of step, the sum stays small. Once they are in step, it grows into a clear wave.
Time is slowed down here.
With 14 cells, the in-step sum is several times bigger. Under a real electrode there are millions of cells, so the difference is far larger.
Why it matters for data analysis
- EEG measures groups, not cells. Every feature you compute describes the shared activity of a large patch of cortex. You can't get single-cell detail from scalp data.
- A bigger signal can mean more synchrony. Power in a frequency band goes up when more neurons work in step, not only when they work harder. Keep this in mind when you read band power.
- Some activity is invisible. Deep structures, small patches and badly aligned cells hardly reach the scalp. Lesson 18 shows what this means for locating sources.
- It is the same maths you know. Sums of in-step signals grow with the number of signals. Sums of out-of-step signals grow only with its square root.
Averaging trials to find a hidden response (lesson 11) uses this rule.
Key terms
| Term | Meaning |
|---|---|
| Neuron | A nerve cell that passes messages with electric currents |
| Action potential | A neuron's short "spike", about 1 ms long |
| Postsynaptic potential | The slower response, tens of milliseconds, where a message arrives |
| Pyramidal neuron | The main cell of the cortex, with a long branch towards the surface |
| Dipole | A pair of opposite charges; here, the current along one neuron |
| Synchrony | Many neurons active at the same time, in step |
Check yourself
- Why does scalp EEG mostly show postsynaptic potentials, not action potentials? Postsynaptic potentials last tens of milliseconds, so many overlap and add up. Spikes last about 1 ms and rarely line up.
- Why do pyramidal neurons matter more than other cells? They point the same way, so their dipoles add up. Cells pointing every way cancel out.
- 100 signals of the same size are added together. About how much bigger is the sum when they are in step, compared with random timing? In step, the sum grows with N (100). With random timing it grows with the square root of N (about 10). So it is about 10 times bigger.
Go further
- Buzsáki G., Anastassiou C. A., Koch C. (2012). The origin of extracellular fields and currents: EEG, ECoG, LFP and spikes. Nature Reviews Neuroscience 13(6): 407–420.
- Tao J. X. et al. (2005). Intracranial EEG substrates of scalp EEG interictal spikes. Epilepsia 46(5): 669–676. The 10 to 20 cm² finding.
- Azevedo F. A. C. et al. (2009). Equal numbers of neuronal and nonneuronal cells make the human brain an isometrically scaled-up primate brain. Journal of Comparative Neurology 513(5): 532–541. The neuron counts.
For learning only, not medical advice.
Previous: What EEG is, and what it isn't · Next: From brain to file: the signal path (coming soon)