Where the EEG signal comes from

What does this show?

Notice in the third panel that the number of cells did not change between the two traces. Only the timing changed. When cells work together their signals add up; when each works at a different time they cancel each other out.

Why do I need this step?

Every decision you make later in EEGLAB depends on knowing where this signal comes from: why you pick a certain reference, why ICA can separate blinks from brain activity, and why electrode Cz does not tell you exactly what happens right under it.

The idea in two lines: an electrode does not hear one cell. It hears the sum of millions of cells working together at the same time, plus any other electrical signal from the body or the room.

How do I do it in EEGLAB?

There are no EEGLAB steps in this lesson, because it explains what the signal measures before you open it.

You will meet this idea in the GUI later in three places: when you first scroll through the signal in microvolts in the GUI lesson, when you choose a reference in the re-referencing lesson, and when you read ICA components that separate brain activity from blinks and muscle.

How do I write it as code?

This script is optional. Run it after the MATLAB basics lesson if you want to see the idea in numbers. It does not need EEGLAB.

The script adds up 1000 small sources, each oscillating at 10 Hz. The first time, each source starts at a random time. The second time, they all start together.

% Sum of many small sources: random timing vs. synchronized timing
nSources = 1000;              % number of small sources
srate    = 128;               % samples per second
t        = 0:1/srate:2;       % 2 seconds of time
f        = 10;                % frequency in Hz (alpha range)

% Each source gets its own random phase (its own timing)
phases   = 2*pi*rand(nSources, 1);
asyncSum = sum(sin(2*pi*f*t + phases), 1);

% All sources share the same phase
syncSum  = sum(sin(2*pi*f*t + zeros(nSources, 1)), 1);

fprintf('Peak of random-timing sum: %.0f\n', max(abs(asyncSum)));
fprintf('Peak of synchronized sum:  %.0f\n', max(abs(syncSum)));

figure;
plot(t, syncSum); hold on;
plot(t, asyncSum);
xlabel('Time (s)'); ylabel('Summed amplitude (a.u.)');
legend('Synchronized', 'Random timing');

• The peak of the synchronized sum is 1000, exactly the number of sources.
• The peak of the random-timing sum is much smaller, usually between 10 and 50, about 28 on average. The number changes on every run because the phases are random, but it stays close to the square root of the number of sources, not the number itself.
• Simplification: real cells do not oscillate as pure sine waves, and the skull weakens and spreads the signal. But the rule “synchrony makes the signal bigger” holds.

Which value should I choose?

There is no value to choose in this lesson. Instead, this table sums up what reaches the electrode and what does not. Come back to it when you first look at the signal.

Everything in the table that is not brain activity is called an artifact: a non-brain signal mixed into the recording. Stations 3 and 4 are about removing it.

How do I check it worked?

Answer each question in your own words before reading the answer after it.

• Why does EEG not usually pick up action potentials? Answer: they are very short, so they do not line up across many cells precisely enough for their sum to reach the scalp.
• What two conditions make pyramidal cell signals add up? Answer: the cells must be parallel and point the same way, and they must be active at the same time.
• A very large wave lasting about half a second appears at Fp1. Is it most likely from the brain? Answer: no. Fp1 is above the left eye, and a large, short wave there is a typical sign of an eye blink. You will learn to identify it in station 4.
• Does electrode Oz measure the visual cortex alone? Answer: no. Current spreads before it reaches the scalp, so the electrode can pick up activity from distant regions.

What mistakes do people make?

• Thinking EEG records neurons “firing”. What reaches the scalp is mainly postsynaptic potentials, not action potentials. So EEG does not directly measure how often cells fire.
• Thinking each electrode measures only the brain under it. Current spreads through the brain, fluids and skull before it reaches the scalp. So an electrode can pick up activity from a distant region (Luck, Appendix 1). Electrode Oz does not mean “visual cortex only”.
• Thinking everything you see is brain activity. Blinks, muscles and power lines appear in the recording, and are sometimes much larger than the brain signal.
• Thinking the voltage at an electrode is absolute. Every channel is the voltage difference between an electrode and a reference electrode. There is no place on the body with zero voltage, so the reference affects every channel (Luck, Appendix 1). This is the topic of the re-referencing lesson.
• Thinking a bigger signal means more activity. Amplitude also depends on synchrony. The same number of cells can give a large signal when synchronized and a small one when not (Cohen 2017).

Going further

• EEGLAB Concepts guide: https://eeglab.org/tutorials/ConceptsGuide/LearningEEG.html (Where to learn EEG)
• Luck, Applied Event-Related Potential Data Analysis (1st edition): https://socialsci.libretexts.org/Bookshelves/Psychology/Biological_Psychology/Applied_Event-Related_Potential_Data_Analysis_(Luck)/12%3A_Appendix_1%3A_A_Very_Brief_Introduction_to_EEG_and_ERPs (Appendix 1: A Very Brief Introduction to EEG and ERPs > The EEG)
• EEGLAB introduction part 2: The origin of the EEG signal: https://www.youtube.com/watch?v=KhEKRHw0qFg (Full video)
• Cohen (2017), Where does EEG come from and what does it mean? Trends in Neurosciences 40(4): https://doi.org/10.1016/j.tins.2017.02.004 (Whole article)