[Eeglablist] (no subject)

[giulio balestrieri]

Hi,

I'm a master student in computational neuroscience and I have just started
to work with EEG data.
I'd like to know why an EEG recording should have at least 30* (the number
of channels) 2 data samples to undergo ICA decomposition. I do understand
that the more channels are included in the analysis the more data
points I need for a robust ICA, but I'd like to know why precisely this
ratio (or how have you established it).
Thanks in advance for your help (and for your great work!).

Best,
Giulio.